Deformation characteristics of a 38 m deepexcavation in soft clay

Guo B. Liu, Rebecca J. Jiang, Charles W.W. Ng, and Y. Hong

Abstract: To meet the increasing demand for underground space for economical development and infrastructural needs,more and more deep excavations have been constructed in Shanghai. In this paper, field performance of a 38 m deep multi-strutted excavation in Shanghai soft clay is reported. The deep excavation was retained by a 65 m deep diaphragm wall. In-clinometers as well as settlement and heave markers were installed to monitor the performance of the deep excavation. Thisproject provides an unusual opportunity to study the differential heaves of center columns and diaphragm walls during exca-vation. Because of the significant stress relief resulting from the 38 m deep excavation, maximum heaves of the center col-umn and diaphragm wall panel were about 30 and 16 mm, respectively. The measured ratio dp/H (heave/final excavationdepth) of column is less than 0.1% whereas the observed dp/H of the diaphragm wall panel is about 0.04%. The maximumdistortion between the column and the diaphragm wall panel is smaller than 1/500, which is within the limit range proposedby Bjerrum in 1963. Owing to careful construction control, stiff strutting system, and compaction grouting, the measuredlateral wall deflections and ground settlements at this site are generally smaller than other shallower excavations in soft claysin Shanghai, Singapore, and Taipei.

Key words: multistrutted excavations, soft clays, field data.

Résumé : Afin de répondre à la demande croissante d’espace souterrain pour le développement économique et les besoinsen infrastructures, un grand nombre d’excavations profondes ont été réalisées à Shanghai. Dans cet article, la performanced’une excavation à entretoisement multiple de 38 m de profondeur dans l’argile molle de Shanghai est discutée. L’excavationprofonde a été retenue par un mur diaphragme de 65 m de profondeur. Des inclinomètres et des marqueurs de consolidationet de soulèvement ont été installés pour suivre la performance de l’excavation profonde. Ce projet offre une opportunité par-ticulière d’étudier les soulèvements différentiels des colonnes du centre et du mur diaphragme durant l’excavation. Les sou-lèvements maximums de la colonne du centre et du panneau du mur diaphragme étaient d’environ 30 et 16 mmrespectivement, en raison du relâchement significatif des contraintes causé par l’excavation de 38 m de profondeur. Le ratiodp/H (soulèvement/profondeur finale de l’excavation) mesuré de la colonne est de moins de 0,1 %, tandis que dp/H du pan-neau du mur diaphragme est d’environ 0,04 %. La distorsion maximale entre la colonne et le panneau du mur diaphragmeest inférieure à 1/500, ce qui est à l’intérieur des limites proposées par Bjerrum en 1963. Grâce à un bon contrôle de laconstruction, d’un système d’entretoisement rigide et de scellement en compaction, les déflections latérales des murs et lestassements du sol sur ce site sont généralement inférieurs à d’autres excavations moins profondes dans l’argile molle à Shan-ghai, Singapore et Taipei.

Mots‐clés : excavations à entretoisement multiple, argiles molles, données de terrain.

[Traduit par la Rédaction]

Introduction

The number of multistrutted excavations constructed inmany major cities around the world has been increasing inrecent decades. For the construction of underground metrostations, the excavations are generally long and narrow. Mon-itoring of soil deformation around excavations and internalforces of retained structures is vital and useful for geotechni-cal designers and engineers to verify design assumptions andreduce construction risk during the excavation process (Whit-tle et al. 1993; Ng 1998; Ou et al. 1998; Long 2001; Finno etal. 2002; Ng et al. 2004; Leung and Ng 2007). Case histories

on field performance and numerical analyses of excavationsin soft clays have been reported by many researchers. Whittleet al. (1993), Finno et al. (2002, 2007), and Hashash andWhittle (1996) described field data and numerical simula-tions on excavations in Boston Blue clay (BBC) in the US.Nicholson (1987), Vuillemin and Wong (1991), and Wallaceet al. (1993) introduced the field performance of deep exca-vations retained by diaphragm walls or sheet piles in Singa-pore soft clay. They showed that the ratios of maximumlateral wall deflection (dhm) and ground settlement (dvm) toexcavation depth (H) typically range from 0.43% to 0.93%and from 0.67% to 1.5%, respectively. Ou et al. (1993) and

Received 10 May 2011. Accepted 9 August 2011. Published at www.nrcresearchpress.com/cgj on 22 November 2011.

G.B. Liu and R.J. Jiang. Department of Geotechnical Engineering, Tongji University, Shanghai, China.C.W.W. Ng and Y. Hong. Hong Kong University of Science and Technology, Department of Civil and Environmental Engineering, ClearWater Bay, Hong Kong.

Corresponding author: Y. Hong (e-mail: chrishy@ust.hk).

1817

Can. Geotech. J. 48: 1817–1828 (2011) doi:10.1139/T11-075 Published by NRC Research Press

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

anya

ng T

echn

olog

ical

Uni

vers

ity (

NT

U)

on 1

2/04

/11

For

pers

onal

use

onl

y.

Wong and Patron (1993) studied some deep excavations suchas the Taipei National Enterprise Center (TNEC), TaiwanTax and Formosa excavations in Taipei soft clay. They re-ported that measured dhm/H and dvm/H values in this softclay typically vary from 0.32% to 0.9% and from 0.23% to0.6%, respectively. Each excavation had a high factor ofsafety against basal heave (FOSbase). Liu et al. (2005) re-ported the performance of a 15.5 m deep multistrutted exca-vation for a metro station in Shanghai soft clays. Relativelysmall wall deflections and ground settlements were measuredas compared with similar case histories worldwide. No sig-nificant “creep” deflection of the diaphragm wall could beidentified over a 60 day concrete curing period. Continuousground settlements accompanied by the dissipation of porewater pressures were observed. Wang et al. (2005) reportedthe characteristics of wall deflections and surface ground set-tlements at six deep multistrutted excavations in Shanghaisoft soils. The ratio between the measured maximum wall de-flection and the depth of the excavation (dhm/H) in Shanghaiwas less than 0.7%, which was similar to the ratio measuredin Taipei, but it was substantially smaller than Peck’s bound-ing limit of 1%. At each station, the measured maximum dis-placement was less than 0.45% of the final excavation depth.The measured maximum settlements of the six metro excava-tions fell within zone I of Peck’s classical normalizedsettlement–distance chart. The observed relatively small max-imum wall deflections were likely attributed to the use ofprestressed struts in which the stresses were constantly ad-justed to about 0.7 times the total vertical stress during theexcavation and the short horizontal span of excavation.Shanghai is located in the typical soft soil area in east-

south China. Many deep multistrutted metro excavationshave been constructed in this congested area in recent years.In fact, the design of these deep excavations is generallybased on semiempirical approaches. Therefore, field monitor-ing is essential for back-analysis of deep excavations inShanghai. The metro excavation described here is 38 m deepin Shanghai. As far as the authors are aware, case historieson deformation of structures and ground settlements for sucha deep excavation in soft clay are very limited. In this study,heave of three columns in the middle of excavation duringthe excavation process has been recorded. This offers an un-usual opportunity to study the heave characteristics of thecolumns as a result of stress relief. These data provide an im-portant chance to verify any predictive method for the design ofdeep excavations in future. This paper focuses on the observedperformance of lateral diaphragm wall deflections, the groundsettlements, and heave of columns and diaphragm wall panel.

Geology and soil parametersTypical ground strata are thick soft soils comprising qua-

ternary alluvial and marine deposits in this project. Highwater content, low shear strength, high compressibility, andlow ground bearing capacity are the typical characteristics ofShanghai soft clay (Gao et al. 1986). To assess the soil prop-erties, field geotechnical engineering investigation is vital andnecessary.A field geology investigation has been carried out. Figure 1

shows a typical soil profile at this excavation site. The strataconcerned in this project site consist of seven layers with to-

tal depth up to 70 m. The first upper layer is 1.6 m thick ar-tificial fill. Below it is the soft to medium clay layer with adepth of 33 m, which was deposited in the Holocene andMiddle Pleistocene geological period. A 30.4 m thick densefine sand layer underlies the clay, which was the MiddlePleistocene deposit. Below this stratum is a thick depositionof Early Pleistocene fine sand, which was not penetrated.Average geotechnical parameters of the investigated layers

are also included in Fig. 1. The unloading modulus (E) is ob-tained by laboratory multistage unloading tests on the undis-turbed soil using an oedometer. The unloading stress range ofthe soil samples was varied to suit the anticipated stresschange. For the silty clay below the fill and very soft claythe unloading range was 200–25 kPa. For the sandy silt, siltyfine sand, and fine sand, the unloading ranges were 300–25,400–25, and 500–25 kPa, respectively. Unloading was car-ried out incrementally, with a reduction of 25 kPa for eachstep. When the deformation of the soil specimen remainedunchanged at the stress state, it was unloaded to the nextstress level. The unloading deformation of fine sand is smalland the E value of this stratum is about 200 MPa (not shownin Fig. 1). The main soil strata studied was up to 40 m belowground level. Apart from the unloading modulus, no other in-formation about fine sand is available at this site. More pa-rameters of fine sand in Shanghai were reported by Xu et al.(2003), as shown in Fig. 1.Horizontal and vertical soil permeability coefficients were

measured by laboratory constant head tests. Vane shear testsin the field were conducted in the boreholes. Consolidatedundrained triaxial tests were carried out on saturated softclay to obtain undrained shear strength parameters. The SPTN values were obtained by using a standard 63.5 kg hammer.When the cumulative distance of driving is 30 cm, the totalblows of the hammer give the blow count N. In general,groundwater conditions were approximately hydrostatic from0.7 to 1.0 m below ground level.

Site conditionsShanghai is located at the front fringe of the Yangtze River

Delta in China. The site plan of this deep excavation isshown in Fig. 2. The south side of the construction site isbounded by buildings founded on 63 m deep piles whilebuildings on the north side are founded on shallow founda-tions. The excavation is retained by a 65 m deep concrete di-aphragm wall (1.2 m thick).A construction link was built in the middle section of the

excavation, shown in Fig. 2, for transportation of vehiclesand workmen during the project construction. Twenty rowsof concrete struts were constructed in the horizontal plane.Each row consists of nine struts in the vertical plane to pro-vide sufficient system stiffness for this very deep cut. Hori-zontal and vertical strut spacings are about 8 and 4 m,respectively. A steel column was set at each strut center tobear the loads on the struts. Each column was socketed intoa 1.2 m diameter cast-in-place pile by 2 m.For typical deep excavations in Shanghai, the struts and

columns are temporary structures and would be removedafter excavation. However, in this very deep excavation allthese concrete struts were designed to be permanent. Theyare used to provide the underground space for usage.

1818 Can. Geotech. J. Vol. 48, 2011

Published by NRC Research Press

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

anya

ng T

echn

olog

ical

Uni

vers

ity (

NT

U)

on 1

2/04

/11

For

pers

onal

use

onl

y.

Construction sequence of the deep excavationFigure 3 shows a typical cross section (section I–I, Fig. 2)

and geometry of the excavation. A top-down construction se-quence was adopted in this excavation. This 38 m deep exca-vation was supported by 65 m deep diaphragm walls andnine levels of in situ concrete struts. Construction of thisvery deep excavation in a soft soil area is complex and diffi-cult. The construction period was divided into 12 main con-struction stages.Prior to the main excavation stages, it took about three

months to install deep diaphragm walls, piles, columns, com-paction grouting (below the excavation formation and behindthe wall), and to build a construction link. Then one perime-ter reinforced concrete capping beam was constructed alongthe top of the whole diaphragm wall. The beam is 1.2 mwide and 0.6 m in height, which enhanced the total supportsystem stiffness. As a result, horizontal deflection of the dia-phragm wall prior to the construction of the L1 and L2 struts

due to the excavation was suppressed. Many obstructionswere found buried in the upper excavation stratum, mainlyfrom 1 to 6.5 m below the ground surface. It took 29 days(in section I–I) to excavate the top soil and remove obstruc-tions. Then the first and second struts (i.e., L1 and L2) wereconstructed, at 1.7 and 6.6 m below ground level, respec-tively. It should be noted that the inclinometer data of thisproject were collected from the excavation below 6.6 m.This means that the wall deflection analysis was from the be-ginning of excavation in stage 4.Previous experiences in Shanghai show that narrow section

excavation (6 to 8 m width) could effectively reduce wall de-flection and ground settlement (Wang et al. 2005), so thesame construction method was adopted in this project. Thetotal construction duration for stepwise excavation and prop-ping in this project was 155 days. Each concrete strut wasconstructed as rapidly as possible. Generally it took 3–4 days to bind the reinforcing cage of one strut and to com-plete casting and curing of concrete. When the excavationreached the formation level, a 1.3 m thick concrete bottomslab was constructed. Other detailed construction stages aresummarized in Table 1.Dewatering wells were arranged inside the excavation.

When the excavation depth reached the aquifer, dewateringmust be done before further excavation. Dewatering in thisexcavation is summarized in Table 1. From stage 5, the watertable on the excavated side was maintained at about 1 m be-low each incremental level of excavation by dewatering.

Construction of the columns and concrete strutsA hollow steel column was socketed into a cast-in-place

pile by 2 m. Section dimensions of the column were650 mm × 650 mm × 18 mm. At the location of the struts(0.5 m above the designed excavation depth), steels were at-tached crosswise to the hollow column and were joined to thepreformed steel junctions on the diaphragm wall.

Fig. 1. Soil profile and average geotechnical parameters (cu, undrained shear strength; E, unloading modulus; e, void ratio; k, permeabilitycoefficient; kh, horizontal permeability coefficient; kv, vertical permeability coefficient; SPT N, blow numbers of standard penetration test; w,water content; WL, liquid limit; WP, plastic limit; g, bulk density). Parameters of fine sand reported by Xu et al. (2003): w = 23.8%; g =19.5 kN/m3; e = 0.68; plastic index = 10.7.

Fig. 2. Plan view of the site.

Liu et al. 1819

Published by NRC Research Press

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

anya

ng T

echn

olog

ical

Uni

vers

ity (

NT

U)

on 1

2/04

/11

For

pers

onal

use

onl

y.

For such a deep excavation in soft soil, efficient design ofthe entire support system is important for successful con-struction. The strut structural system must have adequatestrength, stiffness, and stability. The vertical stiffness coeffi-cient, EI/h4, was estimated to be in excess of 7.8 MPa. InShanghai, the strut structural system for excavation can be

considered structurally stable if its vertical stiffness coeffi-cient is greater than 2.0 MPa. No prestress was applied oneach concrete strut.

InstrumentationTo monitor the performance of this very deep excavation,

various instruments were installed in the project site. In thispaper only the data taken by the inclinometer, heave marker,and settlement marker are presented (see Fig. 2). Inclinome-ter tubes were fixed to steel reinforcement cages and con-creted in the retained wall. The casing was installed with apair of grooves oriented in the expected direction of move-ment. To avoid twisting of the casing, the grooves werechecked before installation to ensure that its angle of twistdid not exceed 0.1%. The casing was fixed tightly to the steelcages of the retained wall and then the groove direction waschecked after lowering the cages into the bentonite. Measure-ments of the twist angle of grooves were repeated after theconcreting. The allowable value of angle of twist is 0.2%.Measurement probes with a resolution of 0.02 mm over a50 mm gauge length for a temperature range of –20 to+50 °C were used for inclination monitoring.Surface settlement monitoring markers were buried 0.3 m

into soil. They were installed at a spacing interval of 1 to1.5 m perpendicular to the two long sides of the excavation.All settlement monitoring was taken with a leveling instru-ment with a stated standard deviation of ±0.5 mm for onekilometre double run leveling.Column heave markers were set on top of the columns (see

P1 in Fig. 3), which are supported by deep piles, and on topof the corresponding diaphragm wall. The measuring methodused and its accuracy for monitoring heave of columns anddiaphragm walls is similar to that for ground settlement.

Unusual opportunity to observe heave of columnsFor deep excavation in a soft area, columns are usually used

to bear the weight of the struts. Column heave occurs withvertical stress relief in the excavation. If column heave islarge, it will cause differential settlement between the columnand the retaining wall and generate the secondary moment tostruts. For some reason, data on the heave of columns arerarely measured and reported.In this very deep and long multistrutted excavation, nine

concrete struts in the vertical plane were cast and these per-manent concrete struts are large in section and weight. Set-ting the columns is necessary for the strut system. Thisexcavation offers an unusual opportunity to observe the de-velopment of column heave and its effects on the supportsystem.

Summary of 10 other metro excavations inShanghaiDetails of 10 other multistrutted metro excavations in

Shanghai soft soil area are summarized in Table 2. The sup-port system of the excavation is diaphragm walls. The finalexcavation depth ranges from 16.5 to 23 m. Generally, theseunderground station excavations are largely rectangular andapproximately 18–23 m wide. Each diaphragm wall was sup-ported by prestressed steel struts. In each case, the strut is

Fig. 3. Cross section and geometry of the excavation at section I–I.

Table 1. Main stages of construction.

Stage Construction operation Day1 Construct diaphragm wall and cast-in-place piles 742 Construct grouting, vertical column, and

construction link95

3 Excavate to L1, L2; construct L1, L2 struts 1144 Excavate to L3, construct L3 struts 1545 Dewater to 17.0 m, excavate to L4, construct L4

struts170

6 Dewater to 21.0 m, excavate to L5, construct L5struts

185

7 Dewater to 24.7 m, excavate to L6, construct L6struts

201

8 Dewater to 28.7 m, excavate to L7, construct L7struts

218

9 Dewater to 32.7 m, excavate to L8, construct L8struts

236

10 Dewater to 36.1 m, excavate to L9, construct L9struts

255

11 Dewater to 41.0 m, excavate to L10, cast bottomslab

279

12 One month after the construction of bottom slab 309

1820 Can. Geotech. J. Vol. 48, 2011

Published by NRC Research Press

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

anya

ng T

echn

olog

ical

Uni

vers

ity (

NT

U)

on 1

2/04

/11

For

pers

onal

use

onl

y.

16 mm thick and 609 mm in diameter (external) with a hori-zontal spacing of 3 m and average vertical spacing of struts issummarized in Table 2. Each prestressed strut was installed0.3 m above each level of the excavation. The prestressedload was actively adjusted to maintain at about 70% of thetotal vertical stress during excavation period.These metro excavation data are used for comparison. No

monitoring data of the columns were recorded for these exca-vations. Calculation of system stiffness (Ks) and factor ofsafety against basal heave (FOSbase) are explained later.

Observed performance of the excavation

Lateral diaphragm wall deflectionLateral diaphragm wall deflection was obtained from 14

inclinometers (shown in Fig. 2). But only the observed dataof six inclinometers (I1 to I6), which present three typical ex-cavation sections, are analyzed here. I1 and I2 are near thewest corner of the excavation, I3 and I4 are near the middleexcavation plane, and I5 and I6 are at the east part of excava-tion. The profile of lateral wall deflection during the mainexcavation (stages 4–11) are shown in Fig. 4. Because of theconstruction program, inclinometer measurement could notcommence at the start of the excavation. In other words,wall deflections due to construction stages S1 and S2 are notincluded in Fig. 4. The recorded data start after excavating6.6 m below ground surface.The deflection profiles developed into a bulged profile in-

ward as excavation depth increased. The maximum wall de-flection is 54 mm (at I5). The inclinometer values ofdiaphragm walls on the south side (I2, I4, and I6) are smallerthan the ones on the north side (I1, I3, and I5) at the sameexcavation depth.The building on the south side is supported by 63 m long

piles, which transfer surcharge to a deep bearing stratum (i.e.,silty sand). While the building on the north side is foundedon a shallow foundation and hence surcharge is transferredto shallow depth near the ground surface. Therefore, it is notsurprising to find different lateral wall deflections on the twosides.Toe movement of the retaining wall was found to take

place at I3 and I4. When the excavation depth approached28 m (stage 8) below the ground surface, the toe movementsof all six monitored walls increased with excavation depth.

From the various metro excavation experiences in recentyears in Shanghai, there was no evidence to suggest that soilheave contributed significantly to toe movement (Wang et al.2005) as no correspondingly large ground settlement was ob-served. No large ground settlement was observed in this ex-cavation (details to be discussed later) and there was noevidence that significant soil heave happened below the exca-vation level. One factor that may contribute to the toe defor-mation is the ratio of D to H, where D is the embeddedlength of the wall, and H is the final excavation depth. Herethe embedment ratio is 0.7. Compared to the other 10 metroexcavations (range from 0.73 to 0.94), the D/H ratio in thisstudy is the smallest. Larger embedment ratio of the dia-phragm wall could help to suppress toe movement.Another possible reason for the toe deflections at I3 and I4

may be related to the construction link built over the dia-phragm wall at this excavation section. Every work day,heavy transport vehicles passed through the link. This contin-ual dynamic loading may have contributed to toe movement.

Maximum wall deflections and locationThe relationship between the maximum lateral wall deflec-

tion (dhm) and excavation depth in this deep excavation isshown in Fig. 5. Measured results from the 10 metro excava-tions in Shanghai and excavations in Boston, Singapore, andTaipei soft clays are also shown for comparison.Comparisons of geotechnical parameters of soft clays in

the aforementioned four cities indicate that moisture content,effective cohesion, and effective angle of friction of BBC andsoft clays in Shanghai and Taiwan are similar, while watercontent of Singapore soft clay is higher (varies from 50%–90%). The effective cohesion, effective angle of friction, andundrained shear strength of Singapore soft clay are lowerthan those of Shanghai soft clay.The dhm/H ratio of excavations in Shanghai, BBC, and Tai-

wan soft clays are mainly located between two limit referencelines: dhm/H = 0.2% and 0.68%. The dhm/H value in Singa-pore soft clay is scattered with the maximum value up todhm/H = 1.6%, which is larger than that in other soft claysites. Figure 5 also shows that although the excavation depthin this study was larger than that in other case histories, themaximum wall deflection was relatively small.It should be noted the lateral wall deflection of stage 3 was

not recorded in this project (6.6 m deep excavation). The

Table 2. Summary of 10 other metro excavations in Shanghai.

SiteaStrutNo.

Thicknessof wall (m)

Excavationdepth (m)

Average verticalstrut spacing (m)

Systemstiffness, Ks FOSbaseb dhm (mm)

Pudianc 4 0.8 16.3 3.8 626.4 1.81 69Yanchang 4 0.8 15.2 3.6 752.2 1.62 82Yangshupu 4 0.8 16.5 3.8 626.4 1.71 94Tianyueqiao 4 0.8 15.0 3.6 777.6 1.51 42Damuqiao 5 0.8 16.9 3.1 1414.3 1.91 54Pudongc 5 0.8 17.3 3.5 911.3 1.91 49Luban 5 0.8 18.2 3.6 777.6 1.86 72Nanpu 5 1.0 22.6 3.8 1223.4 2.01 110Xizangnanc 7 1.0 23.0 3.2 2432.8 1.71 49Shangtiguan 7 1.0 22.7 3.3 2286.5 1.84 81

aThe support system in these 10 metro excavations are multistrutted.bFactor of safety against basal heave.cData from Wang et al. (2005).

Liu et al. 1821

Published by NRC Research Press

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

anya

ng T

echn

olog

ical

Uni

vers

ity (

NT

U)

on 1

2/04

/11

For

pers

onal

use

onl

y.

summarized lateral wall deflection due to 6.6 m deep excava-tion from these 10 metro excavations in Shanghai rangesfrom 8 to 32 mm. If this range of values is added to themeasured ones, the estimated lateral wall deflection range is62 to 86 mm. The estimated value is plotted and it exceededthe expected value dhm/H = 0.2%.Generally, in Shanghai metro deep excavations, the use of

a multistrutted support system, narrow excavation sections,prestressed steel struts (no prestress was applied in concretestruts in this study), and fast workmanship sequences prob-ably account for the small lateral wall deflection. For this ex-cavation, the presence of compaction grouting below the finalexcavation depth and along both sides of the retaining wall(see Fig. 3) may also help to reduce dhm.

Location of maximum wall deflection with excavationdepthFigure 6 shows the relationship between observed location

of maximum lateral wall deflection and excavation depth. Itcan be seen that the measured location of the maximum lat-eral wall deflection changed as the excavation depth in-creased (from stage 4 to stage 11). For the first 10 mexcavation, all the location of maximum wall deflection fellbelow the excavation bottom. The location of the maximumlateral wall deflection moved upward as excavation depth in-creased. When excavation depth reached 24 m (position ofL7 strut), the location was generally at the excavation bottom.This location was above the excavation bottom when the con-struction was approaching the final excavated level.

Fig. 4. Lateral wall deflections from excavation stage 4 (S4) to stage 11 (S11).

1822 Can. Geotech. J. Vol. 48, 2011

Published by NRC Research Press

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

anya

ng T

echn

olog

ical

Uni

vers

ity (

NT

U)

on 1

2/04

/11

For

pers

onal

use

onl

y.

Liu (1995) and Lee et al. (1998) stated that maximum walldeflection occurred below the excavation level owing to therelatively thick layer of soft clay below the excavation. Thismay partially explain the changes of location of maximumwall deflection. With increasing excavation depth, the shorterdistance to stiffer strata caused the maximum wall deflectionto occur above the excavation level.Previously, excavation depths in Shanghai have generally

been less than 25 m. For comparison, the relationship be-tween location of maximum wall deflection and excavationdepth in a 22.6 m deep excavation (i.e., Xizangnan in Shang-hai) is also included in Fig. 6. It can be seen that variation ofthe 22.6 m deep metro excavation is similar to that in thisstudy if the variation in 6.6 m depth excavation is not takeninto account. More reliable measured data in deep excavationin soft soil is needed to further investigate this relationship indeep excavation in soft clay sites.

Three-dimensional distribution of wall deflectionWall deflection measured along the south and north sides

in this excavation provided an opportunity to study whethercorner effects existed in such a deep and long excavation.Figure 7a shows the relationship between normalized maxi-

mum wall deflection and distance ratio along the north andsouth long sides. The distance ratio is measured from the westcorner to the east corner. Figure 7b shows the maximum lateralwall deflection around the excavation. Three-dimensional dis-tribution of wall deflection could be observed.The support system, the stiffness of the strutting system,

and the ratio of length-to-depth and length-to-width shouldbe taken into account to study the corner effect of the exca-vation. Studies on corner effects in BBC, Singapore, and Tai-wan clays (Liu 1995; Lee et al. 1998; Ou et al. 1998; Finnoet al. 2007) suggested a low ratio of length-to-depth andsmaller ratio of length-to-width give rise to more significantcorner effects. In this project, the ratio of length-to-depth andlength-to-width is 4.6 and 7.6, respectively. Corner effect inthis project was not suppressed by sufficient stiffness of theheavy strutting system.

Maximum lateral wall deflections versus FOSbaseThe maximum lateral wall deflections versus the factor of

safety against basal heave were studied by Mana and Clough(1981) by the statistical collection of many excavations in clayareas around the world (Boston, San Francisco, Chicago, Cali-fornia, Oslo, and others). Long (2001) collected this relation forexcavations in Asia clay sites (Japan, Taipei, Singapore). Manaand Clough (1981) proposed the limit lines for this relationship.Figure 8 shows the maximum lateral wall deflections ver-

sus the factor of safety against basal heave (FOSbase) of thisvery deep excavation and 10 other metro excavations inShanghai. Some excavations in Asian clay collected by Long(2001) were also cited here. Both the measured and estimatedmaximum wall deflection in this study fall below the lowerlimited line, with a factor of safety against basal heave of2.2. The FOSbase values of these 10 metro excavations arewithin the two limit lines (near the lower one).

Fig. 5. Relationship between maximum lateral wall deflection andexcavation depth.

Fig. 6. Relationship between location of maximum lateral wall de-flection and excavation depth.

Fig. 7. (a) Relationship between normalized maximum lateral walldeflection and distance ratio. (b) Maximum lateral wall deflectionaround the excavation.

Liu et al. 1823

Published by NRC Research Press

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

anya

ng T

echn

olog

ical

Uni

vers

ity (

NT

U)

on 1

2/04

/11

For

pers

onal

use

onl

y.

With the increasing FOSbase value there is a decreasingtrend in the dhm/H value for metro excavations in Shanghai.For excavations in Singapore, the dhm/H value is above theupper limited line with greater value of FOSbase. Figure 8also shows that the dhm in Singapore is larger than in Shang-hai with the same excavation depth.The use of concrete struts and thick diaphragm wall may

be the factors influencing the differences excavation inducedmovements in Shanghai and Singapore. Grouting on bothsides of the retaining wall at this site may have also contrib-uted to the smaller dhm/H value than most excavations of the10 reference ones in Shanghai.

Effects of stiffness of support system (KS)Clough et al. (1989) proposed the system stiffness (KS) to

study its effect on the stiffness of the support system on lat-eral wall movement. They identified a relationship betweenKS (eq. [1]) and maximum lateral wall deflection for soft tomedium clay. Shanghai clay is soft to medium so this rela-tionship in this very deep excavation is studied. KS is ex-pressed in the following equation:

½1� KS ¼ EWI

gWh4

where EW is the Young’s modulus of the wall (25 GPa); I isthe second moment of inertia of the wall section, calculatedby I = t3/12 where t is the wall thickness; h is the averagevertical strut spacing of multistrutted support system; and gWis the unit weight of water.Figure 9 shows the relationship between normalized max-

imum lateral wall deflection (dhm/H) and system stiffness ofthis deep excavation and some selected deep excavations ofwhich the depth is larger than 17 m from the 10 metro ex-cavations. dhm/H of this study fell below the line of FOSbasewith small dhm/H value. FOSbase shown in Fig. 9 is calcu-lated by the following equation proposed by Clough et al.(1989):

½2� FOSbase ¼ NcSub

H½g � ðSuu=0:7BÞ�where Nc is the bearing capacity factor, related to length,width, and depth of excavation; Sub is the undrained shearstrength below the bottom of the excavation for excavationdepths ranging from 15 to 23 m, which generally rangesfrom 35–55 kPa (obtained by triaxial test); H is the depthof the excavation; g is the bulk unit weight of the soil abovethe bottom of the excavation; Suu is the undrained shearstrength above the bottom of the excavation, which generallyranges from 35 to 55 kPa for the 10 metro excavations (seeTable 2); and B is the effective width of the excavation,which ranges from 18 to 20.6 m for the 10 metro excava-tions (see Table 2).As shown in Fig. 9, dhm/He values are relatively scattered

at a given Ks, and do not provide a very strong correlationwith Ks. This means that the effect of Ks on dhm/He may notbe obvious. The measured data do not seem to correspondwell with the design curves. The value of FOSbase rangedfrom 1.6 to 2.0, while the dhm/He ranged from 0.3% to0.55%. Similarly, the dhm/He in excavations having a similarKs do not show consistency. For example, for the two excava-

tions with the same FOSbase = 1.9, the excavation with higherKs has a larger dhm/He.

Ground surface settlement

Figure 10 shows ground settlement of this deep excavation.Reported ground settlement from other case histories inShanghai, Singapore, Taiwan, and Hong Kong are also in-cluded for comparison. In this study, the observed maximumground settlement (dvm) was less than 20 mm, with a dvm/Heratio of less than 0.053%. The dvm/He ratio in this study issmaller than those reported from other excavations in Shang-hai clay (i.e., 0.15%) and from excavations in Taiwan softclay (i.e., 0.2%). It is even smaller than the ground settlementin some stiff soil sites, such as decomposed geomaterials (i.e.,completely decomposed granite and completely decomposedtuff) in Hong Kong (Leung and Ng 2007)Considering that there are many underground structures

buried at shallow depths (0 to –6 m) in this project, themeasured ground surface movement at this site may havebeen underestimated due to the strengthening effect of theseunderground services.

Fig. 9. Relationship between normalized maximum lateral wall de-flection and system stiffness.

Fig. 8. Relationship between maximum lateral wall deflection andfactor of safety against basal heave.

1824 Can. Geotech. J. Vol. 48, 2011

Published by NRC Research Press

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

anya

ng T

echn

olog

ical

Uni

vers

ity (

NT

U)

on 1

2/04

/11

For

pers

onal

use

onl

y.

Heave of columnsStudies on heave of columns in deep excavation were lim-

ited by the lack of field monitored data. In this study themonitoring data from three instrumented columns (P1, P2,and P3; see Fig. 2) were obtained and analyzed during exca-vation. This provides an unusual chance to study the charac-teristics of column heave and basal heave. Three parts of thedata analyses are presented here: the development of totalheave with time, the incremental heave at each excavationstage, and the differential column heave and retaining wall.

Development of column heaveWhen an excavation is carried out, soil is removed and

ground heave is anticipated due to stress relief. Figure 11shows the measured cumulative development of heave ofeach column during the main excavation stages. The trendsof these three columns are similar. Larger heave rate is ob-served between stages 6 and 8. After finishing the soil exca-vation, the rate of heave dropped and increases in heavestopped. Maximum heave was about 30 mm.Calculations were made to predict the heave of the soil el-

ement at the center position under the final excavated level.Soil heave can be considered as occurring without change involume of the soil, provided no water is allowed to accumu-late in the excavation. The calculations of immediate heave,therefore, can be made in the same way as for immediate set-tlement. The soil weight is considered as a negative load. Un-loading modulus is used in the calculation. No structure(such as strut, column, and cast-in-place pile) in the soil isconsidered. Because the permeability of the clay is low (asshown in Fig. 1), the following undrained equation given byJanbu et al. (1956) was used to calculate the soil heave:

½3� di ¼ qBIu

Eu

where q is the vertical stress relief inside the excavation; B isthe width of the excavation; Iu is an influence factor given byelastic theory; and Eu is the elastic unloading modulus of thesoil (shown in Fig. 1).As described, the column was socketed 2 m into the cast-

in-place pile (see Fig. 3). Diameter of the pile is 1.2 m andits length is 32 m below the final excavation level. As ex-pected, when soil is excavated, the column is found to heave

upwards. However, the pile suppresses the column heave byfriction force. Calculated soil heave is based on the elastic as-sumption without piles while measured data is taken on thetop of the three columns, which are supported by deep piles.Significant difference between the calculated (without pile ef-fect) and measured heave value (as shown in Fig. 11) is,therefore, not surprising.The effect of the cast-in-place pile (1.2 m in diameter) in

reducing basal heave is obviously dependent on spacing, di-ameter, and length of piles. These factors are considered inthe calculation of soil heave. The 32 m long pile is in thestrata of silty fine sand and fine sand. The friction force, fs,is assumed to distribute uniformly in the soil between thetwo diaphragm wall panels. In situ geology investigationshows that the effective friction angle of the soil is about36.5°, cohesion is zero, the width of excavation is 22.6 m,lateral pile spacing is 8 m, and fs is calculated using

½4� fs ¼ bs 00

where b is a coefficient given by Meyerhof (1976) for the ef-fective friction angle 36.5° (b is near 0.331) and s 0

0 is theaverage effective overburden pressure acting on the pile shaft.Calculated soil heave considering effects of pile is shown

in Fig. 11. It should be pointed out that the calculated heaveis not associated with the construction time. Two main fac-tors are considered. One is the value of the unloading modu-lus (E) and the other is the pile effects. It is worth noting thatthe reported E represents an average value of soil stiffnessduring the entire unloading process. The deduced E is prob-ably smaller than the in situ unloading modulus as stiffnessat small strains was not considered in the measurement.Therefore, an unloading modulus equal to twice the measuredE was also used in the following calculation.Four scenarios were calculated, (i) Eu = E with no pile ef-

fects, (ii) Eu = E with pile effects, (iii) Eu = 2E with no pileeffects, (iv) Eu = 2E with pile effects. The value of E isshown in Fig. 1. The calculated heave results of these fourconditions are plotted in Fig. 11 as reference to the measureddata.In scenario 1, the measured heave of each column is sig-

nificantly overestimated by the calculated heave of soil at thecenter of excavation. In scenarios 2 and 3, which only con-sider either increased soil modulus or pile effects, the calcu-lated column heave overestimates the measured value by 72%

Fig. 10. Relationship between ground settlement and distance be-hind retaining wall.

Fig. 11. Cumulative heave of columns P1, P2, and P3.

Liu et al. 1825

Published by NRC Research Press

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

anya

ng T

echn

olog

ical

Uni

vers

ity (

NT

U)

on 1

2/04

/11

For

pers

onal

use

onl

y.

and 24%, respectively. In scenario 4, in which both pile ef-fects and increased soil modulus are considered, the calcu-lated heave (i.e., 24.9 mm) is close to the measured value (i.e.,29.1 mm). Comparisons between calculated basal heave inscenario 4 (Eu = 2E with pile effects) with that in scenario3 (Eu = 2E with no pile effects) may suggest that the basalheave was reduced by about 42% due to the presence ofpiles installed below the excavation.The preceding parametric study indicates that the presence

of piles below the final excavation level and unloading soilmodulus value on heave are two significant factors influenc-ing basal heave.The deadweight (including self-weight of pile, column,

and props) acting on each pile is about 1843 kN, which isabout 36% of the vertical load required to fully mobilize shaftfriction between the pile and its surrounding soil (about5054 kN as estimated by eq. [4]). Therefore, the influence ofdeadweight acting on the pile is not considered to be signifi-cant.

Incremental heave in excavation stagesThe relationship between incremental column heave and

cumulative vertical stress relief from stage 4 to stage 11 isshown in Fig. 12. Calculated data of soil heave in scenario 1and scenario 2 are also shown for comparison.The response of observed column heave to soil unloading

is varied in each excavation stage. Even with the same verti-cal stress relief (such as stages 7 and 8), the incrementalheave is different. Larger measured incremental heave tookplace from stage 7 to stage 8 in sandy silt (see Fig. 3). Evennegative incremental heave was found in stage 11. The incre-mental column heave P1 in stages 8 and 9 are 10.2 and7 mm, respectively, with a cumulative heave value of30 mm. More than half of the total heave was developed inthese two construction stages. The reason for this may bethat plastic soil deformation occurs underneath the formationlevel when excavation depth exceeds 24 m below ground sur-face. In excavation stage 11 (casting of l.3 m thick bottomslab), the incremental heave of column P1, P2, and P3 isless than 1.5 mm.Calculated results show larger incremental heave occurred

from stage 3 to stage 5 and the calculated incremental valuedecreased with excavation depth. This difference between thecalculated values and measured data may be because the cal-culation is based on elastic theory and plastic deformation ofsoil is not considered.

Magnitude of column heaveFigure 13 shows spatial distributions of column heave and

diaphragm wall from excavation stage 4 to stage 11. It can beseen that the wall heave value is less significant than that ofcolumns. dp/H ratios (dp denotes the maximum heave value)of column and wall are less than 0.1% and 0.04%, respec-tively. The most significant wall and column heave occurredin stages 8 and 9, during which sandy silt was excavated (seeFig. 3). During the same period, lateral wall displacementsmeasured by inclinometers I2 and I6 also increased signifi-cantly compared with previous stages (see Fig. 4). The largeinward wall movement in stages 8 and 9 could have causedsoil on the excavated side to heave upwards. It is worth not-ing that measured unloading modulus (E) of sandy silt is the

largest compared with the strata above (Fig. 1). Therefore,larger column heave measured in stages 8 and 9 is likely tobe mainly induced by inward wall movement, rather than de-formation of sandy silt due to vertical stress relief.At the end of the excavation (stage 11), the column heave

appears to be partially suppressed (see Fig. 13), probably dueto self-weight of the bottom slab acting on the formationlevel.

Distortion of column to diaphragm wallFigure 14 shows the distortion between column and dia-

phragm wall panel from stage 4 to stage 11. The distortionis defined as differential vertical displacement between twopoints divided by the horizontal distance between the twopoints. The distance between P1 to P1-1 and P1 to P1-2 ishalf of excavation width (11.3 m). The same distance is usedfor P2 and P3.It can be seen that the distortion increased with excavation

depth until the completion of the 1.3 m thick bottom slab.From stage 6 to stage 9 the increasing rate was the largest.The observed data shows that both the columns and dia-phragm wall heaved during the main excavation process.Measured maximum wall heave is 15.6 mm (at P1-2 in stage11) and the column heave is 28 mm in stage 11. The soilheave below the final excavation level was suppressed andcame to an equilibrium state with the completion of the bot-tom slab.

Fig. 12. Relationship between incremental column heave and cumu-lative vertical stress relief.

Fig. 13. Spatial distributions of column and wall heave.

1826 Can. Geotech. J. Vol. 48, 2011

Published by NRC Research Press

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

anya

ng T

echn

olog

ical

Uni

vers

ity (

NT

U)

on 1

2/04

/11

For

pers

onal

use

onl

y.

The maximum distortion value is less than 1/500, which iswithin the safe limit (i.e., 1/500) proposed by Bjerrum (1963)based on self-weight settlement. No cracks in the concretestruts were observed during and after this deep excavation.

ConclusionsThe field performance of a 38 m deep multistrutted exca-

vation in soft clay in Shanghai was monitored. Based on thefield observation and comparisons of this deep excavationwith 10 other metro excavations in Shanghai and similar ex-cavations in Singapore and Taiwan, the following conclusionsmay be drawn:

1. As the measurements of lateral wall deflection, dhm, due tothe first 6.6 m excavation was not available from this38 m deep excavation, an estimate of lateral wall deflec-tion due to the first 6.6 m excavation was made from 10other excavations in similar ground conditions and con-struction methods in Shanghai. The sum of normalizedestimated (due to the first 6.6 m) and measured (due tothe last 31.4 m deep excavation) lateral wall deflection,dhm/He, is higher than 0.20% but less than 0.30%, whichis small and falls within the limit set by the Shanghaimetro authority. The small wall deflection may be attrib-uted to the adoption of strict construction control andthe use of compaction grouting.

2. Plane strain behavior is observed for the long and narrowexcavation, except at the two ends, where three-dimensionaleffects are significant.

3. Measured locations of the maximum lateral wall deflectionchanged as the excavation depth increased. For the first24 m excavation, the location of maximum wall deflec-tion fell below the depth of excavation. However, whenthe excavation depth reached 24 m (position of L7 strut)and beyond, the location of the maximum lateral walldeflection shifted upwards gradually, above excavationlevels.

4. With an estimated FOSbase against basal heave to be 2.2for the 38 m deep excavation, the normalized measuredmaximum lateral wall deflection falls below the lowerlimit proposed by Mana and Clough (1981). On the

other hand, however, most of the other case histories inShanghai fall within the two limit lines proposed byMana and Clough (1981). For a given stiffness of sup-port system (Ks), there is a relatively large scatter indhm/He values. This seems to suggest that there is nostrong correlation between measured dhm/He and Ks va-lue in excavations in Shanghai soft clay.

5. Due to the significant stress relief resulting from the 38 mdeep excavation, maximum heaves of center column anddiaphragm wall panel are found to be about 30 and16 mm, respectively. The measured column heave ratiodp/H is less than 0.1%, whereas the observed dp/H of thediaphragm wall panel is about 0.04%. The maximum dis-tortion between the column and the diaphragm wall panelis smaller than 1/500, which is within the limit range pro-posed by Bjerrum (1963).

6. Comparing with theoretical calculations of soil heave dueto the stress relief resulting from the 38 m deep excava-tion, it is deduced that the basal heave was reduced byabout 42% owing to the presence of the piles installed un-derneath of the excavation.

AcknowledgmentsThe authors would like to acknowledge the earmarked re-

search grant 618006 provided by the Research Grants Coun-cil of the Hong Kong Special Administrative Region and tothank many colleagues who have contributed to the reportedfield monitoring work in Shanghai.

ReferencesBjerrum, L. 1963. Allowable settlement of structures. In Proceedings of

the Third European Conference on Soil Mechanics and FoundationEngineering, Weisbaden, Germany. Vol. 2, pp. 135–137.

Clough, G.W., Smith, E.M., and Sweeney, B.P. 1989. Movementcontrol of excavation support systems by iterative design. InProceedings of Current Principles and Practices on Foundationand Engineering, Evanston, Ill., 25–29 June 1989. AmericanSociety of Civil Engineers, New York. Vol. 2, pp. 869–884.

Fernie, R., and Suckling, T. 1996. Simplified approach for estimatinglateral movement of embedded walls in U.K. ground. InProceedings of the International Symposium on GeotechnicalAspects of Underground Construction in Soft Ground, London,15–17 April 1996. A.A. Balkema, Rotterdam, the Netherlands. pp.131–136.

Finno, R.J., Bryson, L.S., and Calvello, M. 2002. Performance of astiff support system in soft clay. Journal of Geotechnical andGeoenvironmental Engineering, 128(8): 660–671. doi:10.1061/(ASCE)1090-0241(2002)128:8(660).

Finno, R.J., Blackburn, J.T., and Roboski, J.F. 2007. Three-dimensional effects for supported excavations in clay. Journal ofGeotechnical and Geoenvironmental Engineering, 133(1): 30–36.doi:10.1061/(ASCE)1090-0241(2007)133:1(30).

Gao, D.Z., Wei, D.D., and Hu, Z.X. 1986. Geotechnical properties ofShanghai soils and engineering applications. In Marine geotech-nology and near-shore /offshore structures. STP 923. AmericanSociety for Testing and Materials, West Conshohocken, Pa. pp.161–177.

Hashash, Y.M.A., and Whittle, A.J. 1996. Ground movementprediction for deep excavations in soft clay. Journal of Geotechni-cal Engineering, 122(6): 474–486. doi:10.1061/(ASCE)0733-9410(1996)122:6(474).

Hulme, T.W., Potter, J., and Shirlaw, N. 1989. Singapore MRT

Fig. 14. Distortion of column and diaphragm wall from stage 4 (S4)to stage 11 (S11). DD, distance between column and diaphragmwall; Dd, differential vertical displacement between column and dia-phragm wall.

Liu et al. 1827

Published by NRC Research Press

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

anya

ng T

echn

olog

ical

Uni

vers

ity (

NT

U)

on 1

2/04

/11

For

pers

onal

use

onl

y.

system. Construction Proceedings, Institution of Civil Engineers,London, 86: 709–770.

Janbu, N., Bjerrum, L., and Kjaernsli, B. 1956. Veiledning vedLøsning av Fundamenteringsoppgaver. NorwegianGeotechnicalInstitute, Oslo, Norway. Publication No. 16. [In Norwegian.]

Lee, F.H., Yong, K.Y., Quan, K.C.N., and Chee, K.T. 1998. Effect ofcorners in strutted excavations: field monitoring and case histories.Journal of Geotechnical and Geoenvironmental Engineering, 124(4):339–348. doi:10.1061/(ASCE)1090-0241(1998)124:4(339).

Leung, E.H.Y., and Ng, C.W.W. 2007. Wall and ground movementsassociated with deep excavations supported by cast in situ wall inmixed ground conditions. Journal of Geotechnical and Geoenvir-onmental Engineering, 133(2): 129–143. doi:10.1061/(ASCE)1090-0241(2007)133:2(129).

Liu, K.X. 1995. Three dimensional analysis of deep excavation in softclay. M.Eng. thesis, National University of Singapore, Singapore.

Liu, G.B., Ng, C.W.W., and Wang, Z.W. 2005. Observed performanceof a deep multi-strutted excavation in Shanghai soft clays. Journalof Geotechnical and Geoenvironmental Engineering, 131(8):1004–1013. doi:10.1061/(ASCE)1090-0241(2005)131:8(1004).

Long, M. 2001. Database for retaining wall and ground movementsdue to deep excavation. Journal of Geotechnical and Geoenviron-mental Engineering, 127(3): 203–224. doi:10.1061/(ASCE)1090-0241(2001)127:3(203).

Mana, A.I., and Clough, G.W. 1981. Prediction of movements forbraced cut in clay. Journal of the Geotechnical EngineeringDivision, ASCE, 107(6): 759–777.

Meyerhof, G.G. 1976. Bearing capacity and settlements of piledfoundations. Journal of the Geotechnical Engineering Division,ASCE, 102(3): 197–228.

Ng, C.W.W. 1998. Observed performance of multipropped excavationin stiff clay. Journal of Geotechnical and GeoenvironmentalEngineering, 124(9): 889–905. doi:10.1061/(ASCE)1090-0241(1998)124:9(889).

Ng, C. W. W., Simons, N., and Menzies, B. 2004. Soil-structureengineering of deep foundations, excavations and tunnels. ThomasTelford, London

Nicholson, D.P. 1987. The design and performance of the retainingwalls at Newton Station. In Proceedings of the Singapore MassRapid Transit Conference, Singapore, 6–9 April 1987. Mass RapidTransit Corp., Singapore. pp. 147–154.

Ou, C.-Y., Hsieh, P.-G., and Chiou, D.-C. 1993. Characteristics ofground surface settlement during excavation. Canadian Geotech-nical Journal, 30(5): 758–767. doi:10.1139/t93-068.

Ou, C.-Y., Liao, J.-T., and Lin, H.-D. 1998. Performance ofdiaphragm wall constructed using top-down method. Journal ofGeotechnical and Geoenvironmental Engineering, 124(9): 798–808. doi:10.1061/(ASCE)1090-0241(1998)124:9(798).

Poh, T.Y., and Wong, I.H. 1998. Effects of construction of diaphragmwall panels on adjacent ground: field trial. Journal of Geotechnicaland Geoenvironmental Engineering, 124(8): 749–756. doi:10.1061/(ASCE)1090-0241(1998)124:8(749).

Tamano, T., Fukui, S., Mizutani, S., Tsuboi, H., and Hisatake, M.1996. Earth and water pressures acting on a braced excavation insoft ground. In Proceedings of the International Symposium onGeotechnical Aspects of Underground Construction in SoftGround, London, 15–17 April 1996. A.A. Balkema, Rotterdam,the Netherlands. pp. 207–212.

Vuillemin, R.J., and Wong, H. 1991. Deep excavation in urbanenvironment: examples. In Proceedings of the 10th EuropeanConference on Soil Mechanics and Foundation Engineering,Florence, Italy, 26–30 May 1991. Balkema, Rotterdam, theNetherlands. Vol. 2, pp. 843–847.

Wallace, J.C., Ho, C.E., and Long, M.M. 1993. Retaining wallbehaviour for a deep basement in Singapore marine clay. InProceedings of the International Conference on RetainingStructures, Cambridge, UK, 20–23 July 1992. Thomas Telford,London. pp. 195–204.

Wang, Z.W., Ng, C.W.W., and Liu, G.B. 2005. Characteristics of walldeflections and ground surface settlements in Shanghai. CanadianGeotechnical Journal, 42(5): 1243–1254. doi:10.1139/t05-056.

Whittle, A.J., Hashash, Y.M.A., and Whitman, R.V. 1993. Analysis ofdeep excavation in Boston. Journal of Geotechnical Engineering,119(1): 69–90. doi:10.1061/(ASCE)0733-9410(1993)119:1(69).

Wong, L.W., and Patron, B.C. 1993. Settlements induced by deepexcavations in Taipei. In Proceedings of the 11th Southeast AsianGeotechnical Conference, Singapore, 4–8 May 1993. pp. 787–791.

Xu, Y., Sun, D., Sun, J., Fu, D., and Dong, P. 2003. Soil disturbanceof Shanghai silty clay during EPB tunnelling. Tunnelling andUnderground Space Technology, 18(5): 537–545. doi:10.1016/S0886-7798(03)00083-X.

1828 Can. Geotech. J. Vol. 48, 2011

Published by NRC Research Press

Can

. Geo

tech

. J. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y N

anya

ng T

echn

olog

ical

Uni

vers

ity (

NT

U)

on 1

2/04

/11

For

pers

onal

use

onl

y.