Você está na página 1de 22

Available online at www.sciencedirect.

com

Computers and Geotechnics 35 (2008) 231–252


www.elsevier.com/locate/compgeo

Deep excavation-induced ground surface movement


characteristics – A numerical investigation
Chungsik Yoo, Dongyeob Lee *

Department of Civil and Environmental Engineering, Sungkyunkwan University, 300 Chun-Chun Dong, Jan-An Gu, Suwon,
Kyong-Gi Do 440-746, Republic of Korea

Received 13 August 2006; received in revised form 5 May 2007; accepted 6 May 2007
Available online 28 June 2007

Abstract

This paper concerns the characterization of deep excavation-induced ground surface movements, using the results of numerical inves-
tigation. A calibrated 2D finite element model using the Lade’s double hardening constitutive model for soil was used to form a database
of the wall and ground surface movements associated with deep excavation. The results indicated that the cantilever and the lateral bulg-
ing excavation stages produce distinctive patterns of ground surface movement profiles, and that final ground surface movement profiles
can be constructed by combining the cantilever and the lateral bulging components with a reasonable degree of accuracy. A two-step
approach for use in the prediction of ground surface movement profiles is proposed.
 2007 Elsevier Ltd. All rights reserved.

Keywords: Deep excavation; Ground movement; Finite element analysis; Building damage; Double hardening model

1. Introduction planning, design, and construction of deep excavations in


urban environments.
Rapid urban developments have resulted in many deep Over the years, there have been a number of studies on
excavation projects for constructions of high-rise buildings the subject of wall and ground movements associated with
and subways. During deep excavation, changes in the state deep excavation. Clough and O’Rourke [5] extended the
of stress in the ground mass around the excavation and work by Peck [20] and developed empirical settlement enve-
subsequent ground losses inevitably occur. These changes lopes. Cording [7] provided a means of estimating the dis-
and ground losses affect the surrounding ground in the tribution of ground movements behind an excavation wall
form of ground movements, which eventually impose direct on the basis of volume relationships based on field observa-
strains onto nearby structures. The magnitude and distri- tions. Ou et al. [19] compiled and analyzed field data
bution of ground movements for a given excavation regarding wall movement associated with deep excavation
depend largely on soil properties, excavation geometry and defined the apparent influence range (AIR) for damage
including depth, width, and length, and types of wall and assessment of adjacent structures. More recently, Yoo [25]
support system, and more importantly construction proce- collected field data on lateral wall movement for walls con-
dures. Because of the increased public concern of the effects structed in soils overlying rock from more than 60 different
of construction-induced ground movements on their prop- excavation sites and analyzed the data with respect to wall
erties, the prediction of ground movements and assessment and support types.
of the damage risk have become an essential part of the The finite element method of analysis has also been
extensively used in studies concerning wall and ground
*
Corresponding author. Tel.: +82 31 290 7644; fax: +82 31 290 7549. movements associated with deep excavation. The studies
E-mail address: ldy0311@skku.edu (D. Lee). by Clough et al. [4], Mana and Clough [17], Wong and

0266-352X/$ - see front matter  2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compgeo.2007.05.002
232 C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252

Broms [23], and Hashash and Whittle [10] were directed


toward the prediction of wall movement for excavations
in soft clay. Cording and O’Rourke [6] and later Cording
[7] developed a scaling relationship based on an elastic
assumption and investigated the effect of relative stiffness
of the wall system with respect to soil on lateral wall
movement.
Most of the aforementioned previous studies focused on
the maximum wall movement. Studies concerning ground
surface movement characterization have been scarce due
in large part to difficulties in obtaining complete sets of
data either from field instrumentation or numerical analy-
sis. Available information on the magnitude and distribu- Fig. 1. Finite element mesh and modeling of interface element for
parametric study.
tion of ground surface movement associated with deep
excavation is somewhat outdated and provides limited
information required for building damage assessment. one-half of the excavation was considered in the finite ele-
This study is directed toward the development of predic- ment model.
tion method for deep excavation-induced ground surface The excavation ground considered was assumed to be
movement profiles that can be used in the framework of composed of a weathered granite soil overlying a weath-
currently available building damage assessment procedures ered rock stratum. The weathered granite soil is the repre-
(Boscardin and Cording [2] and Burland [3]). In order to sentative soil in urban excavation sites in Korea and this
correctly simulate the deep excavation-induced ground type of ground formation is a typical soil profile frequently
movement characteristics, the Lade’s double hardening encountered in Korea. The excavation platform corre-
constitutive model [14–16] was used to describe the sponds to the top of the weathered rock stratum.
stress–strain-strength behavior of the model ground. The Primary variables included the wall bending stiffness
finite element model adopted was validated by available (EI)w, the cantilever excavation depth Hun, the unsup-
data [24] before conducting the parametric study. ported span length L below the lower-most support, and
A parametric study on deep excavation problems soil stiffness Es. Combinations of (EI)w and L generated
encountered in Korea was performed using the validated the range of F shown in Table 1. In addition, a wide range
finite element model to form a database for use in the of conditions was analyzed by varying the primary vari-
development of a prediction method associated with deep ables mentioned above. The parameter F in Table 1 is the
excavation-induced wall and ground surface movements. flexibility ratio, defined by the following equation:
The results of the parametric study were carefully analyzed
so that the ground surface movement characteristics could Es L3
F  ð1Þ
be related to the sources of wall movements. A systematic ðEIÞw
approach for prediction deep excavation-induced ground where Es is the soil stiffness, L is the unsupported excava-
movement profiles was then developed. tion length, and (EI)w is the wall flexural rigidity. The flex-
ibility ratio was originally introduced by Cording and
2. Parametric study O’Rourke [6]. The range in wall flexibility ratio considered
in this study is approximately from 15 to 400, which is con-
A series of 2D finite element analyses were performed sistent with the range for slurry, cast-in-place pile, sheet
using a commercial finite element program ABAQUS [1] pile, and soldier pile walls.
to examine the ground surface movement characteristics
and to form a database for use in the development of a 2.2. Finite element analysis
ground movement profile prediction method. Subsequent
sections discuss the details of the parametric study. 2.2.1. Finite element model
In the finite element modeling, the ground and the wall
2.1. Problem investigated were discretized by using eight-noded plane strain ele-

The configuration of the problem to be analyzed is


shown in Fig. 1, which represents a hypothetical case of Table 1
Conditions analyzed
an excavation. For simplicity, an idealized symmetric plane
strain braced excavation geometry with an excavation F Es (aver) L (m) Hun K (MN/
(MPa) (m) m)
depth H and a width B of 20 and 30 m, respectively, was
considered. The wall is a 25 m in height and has with a 15, 50, 74.2, 120, 150, 176, 26.2, 53.2 3, 4, 5, 1, 2, 3, 15
220, 400 6 4
5 m toe penetration depth at the final excavation stage.
Because of symmetry about the excavation centerline, only Note: EIw, wall flexural rigidity; K, effective axial stiffness.
C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252 233

Eð1mÞ Em
ments, while the struts were modeled by using one-node where, M ¼ ð1þmÞð12mÞ and k ¼ ð1þmÞð12mÞ
spring elements with an effective axial stiffness of 15 MN/ Under the two-dimensional plane strain condition, Eq.
m considering the excavation width. (3) can be reduced to the following equation:
As shown in Fig. 1, a refined finite element mesh extended 2 3
to a depth of 1.0H below the final excavation platform and M k 0
6 7
laterally to a distance of 3.8H from the excavation centerline ½Ci ¼ 4 k M 0 5 ð4Þ
to minimize the effect of the artificial boundaries on the 0 0 G
ground surface movement characteristics.
For the material modeling, the weathered rock stratum Based on the results of a series of preliminary analyses
was assumed to be an elasto-plastic material conforming for model calibration, a relatively low shear modulus of
to the Mohr–Coulomb failure criterion [1]. On the basis 50 kPa with high bulk modulus was assigned to the inter-
of the extensive surveys of urban excavation sites in Korea, face elements.
the mechanical properties for the weathered rock in Table 2 In simulating the step-by-step excavation process, the
were used in the parametric study. On the other hand, the initial vertical state of stress was first created by turning
weathered soil was assumed to follow the Lade’s double on the gravity with the assumption of wished-in-placed
hardening model. Brief discussions on the double harden- wall. The lateral stress state was then created by multiply-
ing model and the model parameters used in the analysis ing the vertical stresses by the lateral earth pressure coeffi-
are given in Section 2.2.2. cient K0 = 0.5. The excavation process was then modeled
A series of preliminary analyses indicated that the inter- by adding and removing elements at corresponding steps.
face modeling of the excavation side is crucial for obtaining Although the ground water lowering due to excavation
realistic ground surface movement profiles associated with generally affects the ground movement characteristics, the
deep excavation. Although ABAQUS provides a surface- ground water was not considered for simplification and
based interface modeling option using ‘contact pair’, the the free-draining characteristics of typical weathered soils
contact pair was not adopted in modeling the interface in in Korea.
this study as significant numerical instabilities were
encountered during the excavation modeling. For that rea- 2.2.2. Constitutive modeling of weathered soil
son, the Desai-type thin-layer interface model [9] shown the It has been shown that a traditional elasto-perfectly-
inset in Fig. 1 was implemented in ABAQUS using UMAT plastic model such as Mohr–Coulomb model does not
[1] to model the interface behavior between the wall facing yield satisfactory results in estimating deep excavation-
and the soil. Details of the Desai-type thin-layer interface induced ground movement, especially for ground surface
model can be found in [9]. settlements [21]. The double hardening model, which has
The stiffness matrix of thin-layer element is the same as been proven to be applicable for the weathered soil fre-
general solid element assuming a linear elastic behavior. quently encountered in Korea [13], was selected and imple-
Therefore, the elastic constitutive matrix of the thin-layer mented in ABAQUS using the user subroutine capability
element is expressed as: to represent the soil behavior in this study. In this section,
  the brief descriptions of the double hardening model based
½C nn i ½C ns i on the nonlinear elasto-plastic model are given as below.
½Ci ¼ ð2Þ
½C sn i ½C ss i Details of the double hardening model can be found in
where [Cnn] and [Css] are related to normal and shear com- [14–16].
ponents, respectively, and [Cns], [Csn] represent coupling ef- In the double hardening model, the incremental total
fects. However, the coupling effects are not included for strains are assumed to be divisible into elastic strains, plas-
simplification in this study. tic collapse strains, and plastic expansive strains. The elas-
For isotropic linear elastic behavior, [C]i can be written tic strains are calculated from Hooke’s law using the
in matrix form as: unloading–reloading modulus defined as:
2 3 Eur ¼ K ur  P a  ðr3 =P a Þn ð5Þ
M k k 0 0 0
6 k M k 0 0 07
6 7 where Kur and n are dimensionless model parameters and
6 7
6 k k M 0 0 07 Pa is atmospheric pressure to make conversion from one
6
½Ci ¼ 6 7 ð3Þ
7 system of units to another more convenient. Thus, the units
6 0 0 0 G 0 07
6 7 of Eur and r3 are the same as units of Pa.
4 0 0 0 0 G 05
The plastic collapse strains are associated with volumet-
0 0 0 0 0 G ric strains and mean effective stress. They are computed
using a cap type yield surface conforming associated flow
Table 2
Typical mechanical properties for weathered rock in Korea rule and a work-hardening relationship which can be deter-
mined from an isotropic compression test. The yield crite-
Material C (kPa) / () w () m Es (MPa)
rion, fc, and the plastic potential function, gc, are expressed
Weathered rock 100 46 6 0.3 500
in terms of the first and second invariants, I1, I2, as follows:
234 C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252

fc ¼ gc ¼ I 21 þ 2I 2 ð6Þ where a and b represent the intercept and slope of a


straight line, respectively.
where I1 and I2 are first and second stress invariants.
From Eq. (12), the increment in plastic expansive work
An empirical relationship for the collapse work-harden-
can be expressed as follows:
ing as a function of fc is defined as:
 p dfp 1
fc dW p ¼   ð15Þ
W c ¼ C  Pa  2 ð7Þ fp 1
b
Pa qW p

where C and p are material parameters and the incremental In implementing the double hardening model using the
plastic collapse work (dWc) can be determined from the user subroutine capability the soil is assumed to follow
derivative of Wc with regard to fc: an elastic behavior and then the stress invariants are calcu-
 1p lated to estimate the current collapsive yield stress and
Cp P 2a expansive yield stress. The current collapsive yield stress
dW c ¼ dfc ð8Þ
P a fc and expansive yield stress are then compared to the maxi-
The plastic expansive strains are related to the deviatoric mum past collapsive yield stress and the maximum past
stresses and they are computed using a conical yield surface expansive yield stress to determine the loading conditions.
and conform non-associated flow rule. The conical yield After comparing the present and the past yield stresses, the
surface is described in terms of the first and third stress maximum past yield stresses are replaced with larger one
invariants, I1 and I3: and then they are stored in the solution-dependent state
 3  m variables array. Four different conditions may exist: (1)
I I1 only elastic strains occur, (2) only plastic collapsive surface
fp ¼ 1  27 6 g1 ð9Þ
I3 Pa is activated, (3) only plastic expansive surface is activated
where m is a model parameter describing the curvature of and (4) both plastic collapsive and expansive surfaces are
the failure surface, and g1 is a work-hardening parameters activated. After determining the loading conditions, the
which defines the size of the failure surface. The plastic po- stresses are calculated by using stiffness matrix suitable
tential function is modeled on the yield function as follows: for each yielding condition and then they are updated
  m  and stored to the stress array.
Pa A total of 14 parameters are required in the double
gp ¼ I 31  27 þ g2   I3 ð10Þ
I1 hardening model to define soil behavior: three parameters
where g2 can be modeled by the following simple (Kur, n, and m) define the elastic behavior, two parameters
expression: (p, c) define the collapsive plastic strains, two parameters
 1=2 (g1, m) define failure surface, and three parameters (S, R,
r3 and t) define the direction strain increment but the required
g2 ¼ Sfp þ t þ R  ð11Þ
Pa parameters can be entirely derived from the conventional
triaxial test with volume change measurement. Table 3 pre-
where S, t, and R are model parameters.
sents the double hardening model parameters of the weath-
An empirical relationship between the plastic expansive
ered soil used in the model ground. Note that the values are
work done (Wp) and fp is defined by the following
based on the previous study [13] which reports ranges of
expression:
 1=q
Wp
fp ¼ a ebW p ð12Þ
Pa
 1=q Table 3
with a ¼ g1 Weppeak
Pa
and b ¼ qW 1 Parameters for the weathered soil
ppeak

where Wppeak and q are constants for a given value of r3 Parameters Weathered soil Strain component
and e is the base for natural logarithms. Wppeak is the value Kur 2:57 Elastic component
of Wp at the peak point and its variation with r3 can be n 0.865
approximately expressed by the following empirical Poisson’s ratio 0.25
relationship: Collapse const. (C) 0.0086 Plastic collapse component
 l Collapse expo. (p) 0.789
W ppeak r3
¼P ð13Þ Yield const. (g1) 38 Plastic expansive component
Pa Pa Yield expo. (m) 0.113
where P is the value of Wppeak/Pa when Pa = 1. The varia- P1. potent. const. (R) 1.406
P1. potent. const. (t) 0.488
tion of the parameter q with r3 can be represented by a sim- P1. potent. const. (S) 0.406
ple expression as follows: Work-hard. const. (a) 1.254
  Work-hard. const. (b) 0.072
r3 Work-hard. const. (P) 0.52
q¼aþb ð14Þ
Pa Work-hard. expo. (I) 0.653
C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252 235

the double hardening model parameters for weathered soils Table 5


encountered in Korea. Material properties for an urban excavation site in Korea
Material C (aver) / w m Es (aver) USCS
2.2.3. Model validation (kPa) (aver) (aver) (aver) (MPa)
() ()
The finite element model adopted in this study was vali-
dated to a limited extent against measured data for an Fill 10 30 8 0.3 17 SM
Weathered 25 40 6 0.3 30 SM
urban excavation site in Korea [24]. The same modeling soil
approach for parametric study given in Section 2.2.1 was Weathered 300 41 6 0.3 110 None
employed in the finite element modeling. Fig. 2 presents rock
the standard cross-section and soil profile for the site. The
wall of 26 m in height with an embedment depth of approx-
in terms of the Mohr–Coulomb shear strength parameters
imately 2 m consisted of soldier pile was supported by struts
and the elastic stiffness reported in [24] to give general
and detailed descriptions of the site are given in Table 4 [24].
information of the soils layer and the rock layer. In mate-
As shown in Fig. 2, the vertical spacing of the struts and
rial modeling, although the fill layer and the weathered soil
the unsupported span length were varied by the construc-
in the site have a bit different shear strength parameters as
tion sequences for simulating the field conditions in detail.
shown in Table 5, these soils were considered to have the
The average of the vertical spacing of the struts and the
same double hardening model parameters. This is justified
unsupported excavation length are 2.5 m and 2–3 m respec-
since the fill layer and the weathered soil are reported to be
tively. Cantilever excavation depth of 3–4.5 m, excavation
similar in nature, i.e., classified as SM according to USCS
depth 23.7 m, and the over excavation depth of 2.5–3 m
[24] and the thickness of the fill layer was less than 1.5 m.
are applied to the analysis.
Such an approach was adopted due to the limited informa-
As illustrated in the figure, the ground consists of a fill
tion available and therefore the validation given in this sec-
material, a weathered soil, and a weathered rock. Most
tion should thus be viewed as qualitative. The rock layer is
of the excavation took place within the weathered soil.
assumed to follow the Mohr–Coulomb yield criterion with
Table 5 summarizes the mechanical properties of the site
the material properties in Table 5. Thus, the site was
assumed to be composed of a weathered soil overlying a
weathered rock.
Fig. 3 presents the comparisons between the measured
and predicted ground surface settlements. The solid lines
represent the predictions from the double hardening model
and the points represent the measured results. Note that
due to the general practice of not measuring horizontal
ground surface settlements during excavation, only the ver-
tical ground settlement data were used for validation. As
shown in the figure, the predicted maximum settlement of

Fig. 2. Cross-section and soil profile for an urban excavation site in


Korea.

Table 4
Descriptions of an urban excavation site in Korea
Excavation depth, H (m) 23.7
Wall type Soldier pile (I = 1.26 · 104 m4) and lagging
wall
Soil type Fill + weathered soil
Support type and average Strut (A = 0.0053 m2), horizontal
spacing spacing = 4.0 m, vertical spacing = 2.5 m
Fig. 3. Comparison between results of proposed method and measured
Flexibility ratio (F) 150–250
data.
236 C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252

0.16%H agrees well with the measured one. The extent of stage were kept constant at Hun = 4 m and L = 5 m,
the settlement zone is also well predicted as 2.5–3.0H. More respectively.
importantly, the results agree fairly well in qualitative As expected, an increase in the wall stiffness (EI)w
terms suggesting that the finite element modeling resulted in the decreases in the wall and ground surface
approach, including the constitutive modeling of the movements. The results are described by more of a step
weathered soils using the double hardening model, is rea- function than a gradual change with the wall stiffness. As
sonably appropriate for use in this study. seen in Fig. 4, as the wall stiffness decreases, the shape of
It should however be noted that a small-strain stiffness the wall displacements changes from the cantilever form
model may yield better results, especially in terms of lateral to the lateral bulging form for wall movements, and the
extent of the settlement profiles, such a model was not horizontal component of the ground surface settlement
adopted in this study due to limited information available profiles tend to become convex up and also the vertical
as to the applicability of the model to typical weathered component yields concave down. In terms of the maximum
soils encountered in Korea. Considering that the double values, the location of the maximum settlement tends to
hardening model is based on the nonlinear isotropic elastic- move away from the edge of excavation as the wall stiffness
ity, the results may be comparable to those based on a increases. This trend indicates that for a given excavation
small-strain stiffness model. However, a further study con- condition, the wall bending stiffness influences not only
cerning the use of a small-strain stiffness model for typical the magnitudes of ground surface movements but also
ground conditions in Korea is warranted. the pattern of the movements. In addition, the location
of maximum settlement greatly varies at the flexibility ratio
3. Wall and ground movement characteristics of F = 120.
Moreover, as seen in Fig. 4b, significant horizontal
The results of the finite element analyses were compiled ground surface displacements are developed, as great as
such that the patterns of wall and ground movements could 100–127% of the vertical settlements in terms of maximum
be related to the parameters comprising the flexibility ratio. values. These results demonstrate the need for considering
Important findings are discussed under the subsequent horizontal ground surface displacements when assessing
subheadings. the risk of damage especially for buildings with small resis-
tance to lateral ground displacements, as noted by Cording
3.1. Effect of wall stiffness et al. [8], although the lateral building strains are signifi-
cantly less than the lateral ground strains.
Fig. 4 illustrates the effect of wall bending stiffness on Presented in Figs. 5 and 6 are the results of four cases
the wall and ground surface movement patterns. In the plot having the same flexibility ratio of F = 120 but different
of the ground surface displacements, two sets of curves are combinations of (EI)w and L and different Hun, one with
shown. For the horizontal displacements, the displacement Hun = 1 m and the other with Hun = 4 m. The differences
toward the excavation wall was taken as positive. Likewise in the wall and ground surface movements between the
the downward movement was taken as positive for the ver- cases thus represent the combined effect of (EI)w and L.
tical settlements. Note that cantilever excavation depth and As seen in these figures, the wall and ground surface dis-
the unsupported span length during the lateral bulging placement profiles tend to significantly vary in shape

Fig. 4. Variation of wall and ground movements with (EI)w.


C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252 237

Fig. 5. Variation of wall and ground movements according to combination of (EI)w and L (Hun = 4 m, F = 120 and Es = 26.2 MPa).

Fig. 6. Variation of wall and ground movements according to combination of (EI)w and L (Hun = 1 m, F = 120 and Es = 26.2 MPa).

depending on the combination of (EI)w and L despite the wall flexibility ratio F alone cannot correctly capture the
same F. wall and ground responses to a given excavation.
For the wall deformations, the maximum displacements
of the wall do not vary in direct proportion to the bending 3.2. Effect of unsupported excavation depth
stiffness. Furthermore, the curve 2 with the larger L and
(EI)w yields greater wall and ground surface movements During an excavation, unsupported excavation is inevi-
than the curve 1 with the smaller L and (EI)w due to the tably carried out both in the cantilever and the lateral bulg-
over excavation. A similar trend is shown in curves 3 and ing stages. The effect of cantilever excavation depth Hun on
4, suggesting that L and over excavation have a greater the wall and ground movements is illustrated in Fig. 7. The
influence on the performance of a given excavation than cantilever excavation depth Hun was controlled in the anal-
(EI)w. For each unsupported excavation depth L (=3, 4, yses by varying the maximum depth of excavation before
5, 6 m), the depth of over excavation is 0.5, 1.5, 0, and installing the top most support but keeping the other vari-
1.0 m, respectively. The trends shown in these figures illus- ables constant. Therefore, the differences in the wall and
trate the dependency of the wall and ground movements on ground movements between the different cases solely reflect
the combined effect of (EI)w and L and suggest that the the effect of Hun.
238 C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252

Fig. 7. Variation of wall and ground movements with Hun (F = 120 and Es = 26.2 MPa).

As seen in Fig. 7, it appears that the wall movements Examples with different Hun used to show the influences
during the cantilever stage became significant when allow- of the unsupported span length L for a given wall stiffness
ing Hun greater than 3 m, resulting in the ground surface are illustrated in Figs. 8 and 9. As shown in these figures,
displacement profiles with large volume losses in close an increase in L from 3 to 6 m results in substantial
proximity to the edge of the excavation. A strict provision increases in the wall and ground surface movements by
on Hun should be placed during the early stages of excava- approximately an order of magnitude.
tion for cases in which buildings and buried utilities are An important observation is that an increase in the
located in close proximity to the edge of excavation. unsupported span length L results in increases in the mag-
In addition, the wall deformations are almost constant nitude of ground surface settlements and tends to modify
with the increase of Hun, and similar results are shown the settlement profiles to more a concave downward shape
for the vertical components. Hence, it is seen that maxi- because of the increased deep-seated movements. Although
mum wall deformation and the vertical displacement of the flexibility due to the increase of the unsupported length
ground settlements are more dependent on the wall bend- L is increased, the wall and ground movements are severely
ing stiffness and the unsupported length L than the cantile- varied by over excavation. Such results shown in these fig-
ver excavation depth Hun. ures demonstrate that the excavation procedure during

Fig. 8. Variation of wall and ground movements with L (Hun = 4 m).


C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252 239

Fig. 9. Variation of wall and ground movements with L (Hun = 1 m).

construction significantly affects the patterns of ground (=26.2, 53.2 MPa) and different Hun. In all cases, the
surface displacement profiles and the wall deformations. unsupported depth during the lateral bulging stage was
kept constant at L = 5 m. Note that Es represents the aver-
3.3. Effect of soil stiffness age soil modulus of the excavation ground calculated using
the hyperbolic relationship given by Janbu [12] in the fol-
The relative stiffness between the excavation wall and lowing equation:
the soil has a significant influence on the magnitude and n
Es ¼ K  P a  ðr3 =P a Þ ð16Þ
distribution of wall and ground movements for a given
excavation. With respect to ground movement prediction, According to Wong and Duncan [22], the parameter Kur
it would be more convenient if cases having the same flex- in Eq. (5), which is the modulus parameter determining
ibility ratio but different combinations of (EI)w and Es Eur, is generally 1.2–3 times greater than K, which is the
would yield similar ground surface movement profiles. modulus parameter for Es. Considering this, the average
Figs. 10–15 represent the ground surface movement pro- value of Es was back-calculated from Eur.
files at the cantilever, lateral bulging, and final stages for As seen in these figures, it appears that the cases with
cases having F = 50 and 150 but different combinations a same F would yield similar settlement profiles in terms
of (EI)w (=21.8, 44.4, 65.4, 133.1 MN m2/m) and Es of magnitude and slope despite the differences in the

Fig. 10. Ground movements profiles with soil stiffness (Hun = 4 m): (a) F = 50 and (b) F = 150.
240 C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252

Fig. 11. Ground movements profiles with soil stiffness (Hun = 4 m): (a) F = 50 and (b) F = 150.

Fig. 12. Ground movements profiles with soil stiffness (Hun = 4 m): (a) F = 50 and (b) F = 150.

combination of (EI)w and Es. The general shapes of the hor- be related to the maximum lateral wall movements (dw,max)
izontal displacement profiles for a given F are similar, and for cases having different wall flexibilities. The maximum
the case with less rigid ground and wall tends to yield larger values for each component, i.e., the cantilever and the lat-
movements. An important conclusion is that as long as L is eral bulging, are separately analyzed, and that the results
the same, the flexibility ratio F can be used as a common are only relevant to the cases with ground conditions sim-
index for the prediction of ground surface settlement profiles ilar to those considered in this study.
for cases having different combinations of (EI)w and Es. Presented in Figs. 16–19, dv,max/dw,max and dh,max/dw,max
ratios are plotted against the flexibility ratio F for the can-
3.4. Maximum wall and ground surface displacements tilever as well as the lateral bulging components in a semi-
log plot. The case with the unsupported length L = 5 m
During a design stage, the maximum ground surface set- was chosen to eliminate the effect of over excavation.
tlement (dv,max) is usually taken as a fraction of the maxi- As seen in Fig. 16a, the dv,max/dw,max ratio tends to lin-
mum lateral wall movement (dw,max). The results of the early decrease with increasing F because of the increase
finite element analyses are compiled such that the maxi- of dw,max at an approximately same rate, regardless of the
mum ground surface movements (dv,max and dh,max) can soil stiffness in the cantilever stage. The range of decrease
C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252 241

Fig. 13. Ground movements profiles with soil stiffness (Hun = 1 m): (a) F = 50 and (b) F = 150.

Fig. 14. Ground movements profiles with soil stiffness (Hun = 1 m): (a) F = 50 and (b) F = 150.

is within 20%. In the lateral bulging stage, as the result of dv,max/dw,max ratio. The difference in soil stiffness results
the decrease of the wall bending stiffness, dv,max/dw,max from the lateral expansion of the retained soil.
ratio tends to linearly increase with F, and the rate of As illustrated in Fig. 18, the change of the vertical com-
increase for the soil stiffness remains almost constant ponent lies within 5% because of the decrease in Hun and
within 10%. the immediate installation of struts after excavation. In
The dh,max/dw,max ratio shown in Fig. 17 appears to sig- the lateral bulging stage, the effect of the flexibility is also
nificantly vary with F in the cantilever stage, in the range of insignificant, and only the change associated to the soil
about 30%. This result suggests that even for a rigid wall stiffness appears. As seen in Fig. 19, dh,max/dw,max ratio also
system, a large amount of volume loss can be led in the displays an almost similar trend for the case of Hun = 4 m.
cantilever stage. For a flexible wall system, as a result of However, the range of variation is within 15%, and the
the increase of wall deformation, dh,max/dw,max ratio tends effect related to the wall and the soil stiffness decreases.
to decrease and the effect of soil stiffness is insignificant like This result, as previously mentioned, reflects the influence
dv,max/dw,max. As seen in Fig. 17b, the dh,max/dw,max ratio of Hun. Therefore, dv,max/dw,max is dependent on Hun and
does not appear to significantly vary with F in the lateral F, regardless of the soil stiffness in the cantilever stage,
bulging stage, exhibiting similar results displayed by and the lateral bulging stage is affected by Es, F, and
242 C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252

Fig. 15. Ground movements profiles with soil stiffness (Hun = 1 m): (a) F = 50 and (b) F = 150.

Fig. 16. Variation of dv,max/dw,max with flexibility ratio (Hun = 4 m): (a) cantilever stage and (b) lateral bulging stage.

Hun. As discussed in Section 3.2, the change of the horizon- As seen in the figure, the range of dv,max/dw,max ratio lies
tal component of ground settlement is greatly affected by within 0.5–1.0. This trend is similar to the field data [24] in
the cantilever excavation depth. A constant value for Fig. 21c and strongly supports the trends observed in the
dh,max/dw,max can be assumed for the lateral bulging stage results of the finite element analyses, in which the wall flex-
without introducing a significant error, at least, within ibility increased due to the increase in L as the excavation
the limits of the F considered. proceeded. A similar trend has been reported by O’Rourke
Fig. 20 represents the ratio of settlement volume to wall [18], in which the ratio of the volume of ground settlement
displacement volume (Vs/VL). As seen in Fig. 20, the ratio increased, to as great as 1.0, as the wall flexibility increased
of settlement volume to wall displacement volume almost by the increase in L. This similarity indicates that the way in
linearly increases with increasing F in the semi-log plot. which the ground loss at the wall is reflected to the ground
These results are a direct consequence of the increase in surface depends not only on the dilatancy characteristics
the ground settlement volume relative to the volume of wall of the ground but also on the wall flexibility. The wall flex-
displacement as the wall stiffness decreases. ibility should therefore be taken into consideration when
Fig. 21 illustrates the relationship between the normal- inferring the maximum ground surface settlement from the
ized maximum ground displacements and wall movements. maximum lateral wall movement for a given excavation.
C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252 243

Fig. 17. Variation of dh,max/dw,max with flexibility ratio (Hun = 4 m): (a) cantilever stage and (b) lateral bulging stage.

Fig. 18. Variation of dv,max/dw,max with flexibility ratio (Hun = 1 m): (a) cantilever stage and (b) lateral bulging stage.

4. Ground surface movements characterization cantilever component from the final profile. Comparisons
between the two cases thus provide insights into the general
In the sections that follow, the ground surface move- characteristics of the cantilever and the lateral bulging com-
ment profiles for the conditions analyzed are decomposed ponents of ground surface movement profiles.
into cantilever and lateral bulging components, and the As seen in Fig. 22a for the case of Hun = 1 m, it is evi-
characteristics of each component are discussed. dent that the cantilever components are negligibly small,
and that the final profiles and the lateral bulging compo-
4.1. Components of ground surface movement profiles nents are practically the same. For the case of Hun = 4 m
shown in Fig. 22b, in which significant cantilever-type wall
Illustrated in Figs. 22 and 23 are the ground surface movements are allowed to develop, both the cantilever
movement profiles for two cases, both of which have the and the lateral bulging components are apparent. A sali-
same wall and excavation conditions but different Hun. ent feature observed in Fig. 22 is that the final settlements
Each figure shows three curves; one for the final, and the of the two cases are practically identical, regardless of
other two for the cantilever and the lateral bulging compo- Hun, illustrating that the cantilever excavation depth Hun
nents, respectively. Note that the lateral bulging component has a transitional effect on the final settlements of a given
of a given movement profile is obtained by subtracting the excavation.
244 C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252

Fig. 19. Variation of dh,max/dw,max with flexibility ratio (Hun = 1 m): (a) cantilever stage and (b) lateral bulging stage.

Fig. 20. Relationship between Vs and VL (L = 5 m): (a) Hun = 4 m and (b) Hun = 1 m.

As seen in Fig. 23, cases having different combinations ments for a given excavation depend basically on the
of (EI)w and Es exhibit the same trend. The results illus- depth of cantilever excavation Hun and the relative wall
trated in these figures have significant practical implica- stiffness with respect to soil stiffness.
tions on the prediction of ground movement profiles, as A survey conducted on deep excavation practice in
they imply that the cantilever and the lateral bulging com- Korea as part of this study indicated that the depth of can-
ponents are separable and independent of each other. tilever excavation was approximately 4–5 m with wall flex-
Therefore, any deep excavation-induced ground movement ibility F = 10–250. For this reason, the results for cases
profiles can be conveniently constructed by simply adding having four levels of flexibility ratio, i.e., F = 15, 74.2,
the cantilever and the lateral bulging components. This is 120, and 220 with the soil stiffness of Es = 26.2 MPa are
discussed further later in this paper. presented in Fig. 24 associated with the cantilever-type lat-
eral wall movements.
4.2. Cantilever component As noticed, the vertical settlement profiles are essentially
bilinear in shape with the maximum values occurring near
Cantilever-type movements develop during excavation the edge of excavation. The horizontal displacement pro-
to upper levels of supports. The magnitude and distribu- files change from parabolic to bilinear in shape as the wall
tion of ground movement due to cantilever wall move- bending stiffness decreases. In addition, significant horizon-
C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252 245

Fig. 21. Relationship between dv,max and dw,max: (a) Hun = 4 m, (b) Hun = 1 m and (c) field data.

tal and vertical movements develop in the region within As seen in these figures, the vertical and the horizontal
0.2H from the edge of excavation. Buildings and utilities settlement profiles follow predominantly concave upward
located in close proximity to a deep excavation with a flex- and downward shapes with the maximum values occurring
ible wall, therefore, can experience significant levels of approximately at 0.3–0.4H and 0.5–0.6H away from the
angular as well as lateral distortions even in the early stages edge of excavation for the lateral bulging stage, respectively.
of excavation. Salient features shown in these figures are two-fold.
First, for a given flexibility ratio, the movement profiles
4.3. Lateral bulging component tend to vary depending on the combination of (EI)w and
L with this trend being more pronounced as the wall flexi-
The lateral bulging components of the ground surface bility increases. This trend supports the results presented
movement profiles are illustrated in Figs. 25 and 26 for earlier that L and the depth of the over excavation are by
cases having two levels of flexibility ratio, i.e., F = 15 and far a more important controlling factor of ground move-
220 with different Hun. For each flexibility ratio, a number ment than the wall stiffness (EI)w for excavation cases. Sec-
of profiles having different combinations of (EI)w and L are ondly, although not as apparent, the locations of the
presented. maximum horizontal displacement and vertical settlement
246 C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252

Fig. 22. Ground surface displacement profiles (F = 120 and Es = 26.2 MPa): (a) Hun = 1 m and (b) Hun = 4 m.

Fig. 23. Ground surface displacement profiles (F = 120 and Es = 53.2 MPa): (a) Hun = 1 m and (b) Hun = 4 m.

tend to move slightly farther away from the edge of exca- ferent in magnitude, normalization holds for the ground
vation as L increases, due primarily to the deep-seated surface movement profiles for cases having the same flexi-
movements associated with increasing L. bility ratio. The variation in the region beyond the maxi-
mum values is of little practical importance on account of
4.4. Normalized ground surface displacement profiles the relatively small magnitudes. Although not included
here, other cases exhibited the same trend.
With respect to the ground surface movement predic- Fig. 29 presents normalized profiles for the case of
tion, it would be desirable if normalized relationships can Hun = 4 m with their respective maximum values at the
be established among different cases. Figs. 27 and 28 pres- final excavation stage for the cantilever as well as the lat-
ent the lateral bulging components of the ground surface eral bulging components, which can be readily used to
movement profiles normalized with their respective maxi- make the prediction of deep excavation-induced ground
mum values for cases having F = 15 and 220. movements for cases with similar excavation conditions
Interestingly, the normalized profiles for a given F but considered in this study.
different combinations of (EI)w and L tend to collapse more The normalized profiles for the cantilever component
or less into one curve. This trend indicates that although dif- shown in Fig. 29a represent cases with a cantilever excava-
C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252 247

Fig. 24. Cantilever components of the ground movements profiles (Es = 26.2 MPa): (a) Hun = 4 m and (b) Hun = 1 m.

Fig. 25. Lateral bulging components of ground movements profiles (Hun = 1 m).

tion depth of Hun = 4 m. Fig. 29b presents the normalized respective maximum values obtained either from Fig. 29
lateral bulging components for various flexibility ratios. or local experience. Complete ground surface movement
The flexibility ratios covered in this figure bracket a wide profiles can then be constructed by simply adding the can-
range of cases frequently encountered in practice, and tilever and the lateral bulging components of profiles.
therefore, these profiles can be used to make a prediction
of ground movement profiles for a given excavation with 4.5. Estimation of settlement influence zone
a reasonable degree of accuracy. A two-step approach is
proposed for prediction of deep-excavation-induced The influence zone of ground settlements should be rea-
ground movement profiles, in which the normalized canti- sonably established to evaluate the damage of adjacent
lever and lateral bulging components of ground surface buildings caused by deep excavation. In this study, the
movements are first systematically determined from influence zone of ground movement obtained by FE anal-
Fig. 29a and b, respectively, based on the wall stiffness ysis is compared with that of a previous study by Hsieh and
(EI)w and the flexibility ratio F for a given excavation. Ou [11]. Figs. 30–33 show the normalized ground surface
Actual cantilever and lateral bulging components are then profiles with their maximum horizontal and vertical
obtained by multiplying the normalized profiles with their ground settlements for cases having various flexibility
248 C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252

Fig. 26. Lateral bulging components of ground movements profiles (Hun = 4 m).

Fig. 27. Normalized lateral bulging components of ground movement profiles (Hun = 4 m).

ratios, different cantilever excavation depths, and soil stiff- where dv,max occurs is half the final excavation depth 0.5H.
ness. In these cases, the unsupported excavation span Compared to the results of Hsieh and Ou [11], however,
length is held constant at L = 5 m. In these figures, the dot- dv,max from this study occurs at a shorter distance, approx-
ted line represents the influence zone of ground surface set- imately 0.28–0.36H. In the relationship between dw,max and
tlements obtained by Hsieh and Ou [11] and the solid lines dv,max, the ratio of the distance where dv,max occurs to the
represent the results from this study. depth where dw,max occurs increases to approximately
As indicated in Fig 30a, the ratio of dv/dv,max is approx- 0.51–0.86 for the rigid wall as L increases. This trend indi-
imately equal to 0.5 and agrees well with the results of cates that the distances where dv,max and dw,max occur vary
Hsieh and Ou [11]. The primary influence zone and the sec- with the unsupported excavation length L. For the flexible
ondary influence zone are about to 1.7H and 3H, respec- wall, the ratio increases approximately to 0.47–0.62 with
tively. These results are smaller than those proposed by increasing L.
Hsieh and Ou [11] but agree with the results obtained by The influence zone of the horizontal ground settlements
the method of Clough and O’Rourke [5] as well as the field is examined in this study. In most previous studies, there is
measurements [24] as presented in this study. According to no guideline for the prediction of the horizontal settlement
the studies by Hsieh and Ou [11], the distance from the wall profiles because of the difficulty in measuring horizontal
C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252 249

Fig. 28. Normalized lateral bulging components of ground movement profiles (Hun = 1 m).

Fig. 29. Normalized ground movement profiles (Hun = 4 m and L = 5 m): (a) cantilever component and (b) lateral bulging component.

ground surface displacements, and consequently, only the As shown in Fig. 31 for the case with Hun = 1 m, the results
influence zone for the vertical ground surface profiles has exhibit the same trend as the case with Hun = 4 m except
been proposed. that the displacements at the wall greatly decrease due to
Presented in Fig. 30b is the ratio of dh to dw,max at the the installation of the top most support. As seen in Figs.
wall, being a range of 0.7–1.0. This range decreases with 32 and 33, the results yield similar settlement profiles in
increasing F. For the rigid wall system, the horizontal terms of the magnitudes and the slope despite of the differ-
ground movements at the wall is directly associated with ences in Es.
the wall displacement but the maximum horizontal ground Based on the finite analyses, it can be concluded that the
displacement occurs away from the wall for the flexible primary influence and the secondary influence zones are
wall system due to the lateral bulging. The maximum value approximately equal to 2H and 3H, respectively. These
of dh/dh,max occurs at 0.5H except for rigid wall system, and results agree with the method of Clough and O’Rourke
the primary influence and secondary influence zones for the [5] and almost coincide with the study by Hsieh and Ou
horizontal ground surface movements are approximately [11]. Therefore, the influence zone of ground surface move-
equal to 1.7H and 3.0H, respectively. Such results are sim- ment can be estimated by using the double hardening
ilar to the cases for the vertical ground surface movements. model implemented in this study with a reasonable degree
250 C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252

Fig. 30. Estimation of settlement influence zone (Hun = 4 m, L = 5 m and Es = 26.2 MPa): (a) vertical component and (b) horizontal component.

Fig. 31. Estimation of settlement influence zone (Hun = 1 m, L = 5 and Es = 26.2 MPa): (a) vertical component and (b) horizontal component.

of accuracy, and the double hardening model can be ble hardening model and the finite element model adopted
applied to the damage assessment of adjacent buildings in this study was validated using available field instrumen-
induced by deep excavation in urban areas. tation data. The finite element model was then employed
for a parametric study on deep excavations with emphasis
5. Conclusions on ground movements. On the basis of the parametric
study, a method for predicting deep excavation-induced
This paper presents the results of numerical investiga- ground movement profiles is proposed, which is of prime
tion on deep excavation-induced ground surface move- importance in damage assessment of adjacent structures.
ment characteristics under the ground conditions Based on the results of the present study, the following
encountered in Korea. In order to realistically model conclusions can be drawn:
ground movements associated with deep excavation,
Lade’s double hardening model was incorporated into (1) The general shape of a ground surface settlement pro-
ABAQUS and used to simulate stress–strain behavior of file is closely related to the sources of wall move-
the weathered soil. The appropriateness of the Lade’s dou- ments, and that the unsupported span length has a
C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252 251

Fig. 32. Estimation of settlement influence zone (Hun = 4 m, L = 5 m and Es = 53.2 MPa): (a) vertical component and (b) horizontal component.

Fig. 33. Estimation of settlement influence zone (Hun = 1 m, L = 5 m and Es = 53.2 MPa): (a) vertical component and (b) horizontal component.

significant influence on the magnitude and distribu- by combining the cantilever and the lateral bulging
tion of wall and ground movement characteristics. components.
(2) For a given ground condition, the ratio of the maxi- (4) The primary influence and the secondary influence
mum ground surface settlement to the maximum wall zones are approximately equal to 2H and 3H for ver-
movement decreases with increasing wall flexibility for tical ground movements, respectively. The results
the cantilever component but increase with increasing agree with the previous studies by Clough and
wall flexibility for the lateral bulging components. O’Rourke and Hsieh and Ou.
(3) The cantilever and the lateral bulging stages of exca- (5) The proposed method for predicting ground surface
vation produce distinctive ground surface displace- movement profiles captures the fundamental charac-
ment profiles, which are separable and independent teristics of ground surface movement profiles, and
of each other. Deep excavation-induced ground sur- therefore can be used with a reasonable degree of
face movement profiles for a given excavation can accuracy to make an estimate of ground surface
be predicted with a reasonable degree of accuracy movement profiles.
252 C. Yoo, D. Lee / Computers and Geotechnics 35 (2008) 231–252

References [12] Janbu N. Soil compressibility as determined by oedometer and


triaxial tests. In: Proceedings of the European conference on soil
[1] ABAQUS (ver. 5.8) User’s Manual. Hibbitt, Karlsson & Sorensen mechanics and foundation engineering, vol. 1; 1963. p. 19–25.
Inc.; 1999. [13] Kang Byung Sun, Yang Jae Hyouk. Adoptability of double harden-
[2] Boscardin MD, Cording EJ. Building response to excavation-induced ing constitutive model for compacted decomposed granite soil,
settlement. J Geotech Eng, ASCE 1989;115(1):1–21. KSCE. J Civil Eng 1999;19(3):1013–20.
[3] Burland JB. Assessment of risk of damage to buildings due to [14] Lade PV. Elasto-plastic stress–strain theory for cohesionless soil with
tunneling and excavations. In: Invited special lecture to IS-ToKyo curved-yield surfaces. Int J Soild Struct 1977;13:1019–35.
’95: proceedings of the first international conference on earthquake [15] Lade PV, Nelson RB. Incrementalization procedure for elasto-plastic
geotechnical engineering; 1995. constitutive model with multiple, intersecting yield surfaces. Int J
[4] Clough GW, Hansen LA, Mana AI. Prediction of supported Numer Anal Method Geomech 1984;8:311–23.
excavation movements under marginal stability conditions in clay. [16] Lade PV. Three-dimensional behavior and parameter evaluation of
In: Proceedings of the third international conference on numerical an elastoplastic soil model. Geomech Model Eng Pract 1986:297–311.
methods in geomechanics; 1979. p. 1485–502. [17] Mana AI, Clough GW. Prediction of movement for braced cuts in
[5] Clough GW, O’Rourke TD. Construction induced movements of clay. J Geotech Eng, ASCE 1981;107(6):759–77.
insitu walls. In: Proceedings of the design and performance of earth [18] O’Rourke TD. A study of two braced excavation in sands and
retaining structures, ASCE special conference; 1990. p. 439–70. interbedded stiff clay. PhD thesis, University of Illinois at Urbana-
[6] Cording EJ, O’Rouke TD. Excavation, ground movements, and their Champaign; 1975.
influence on buildings, protection of structures adjacent to braced [19] Ou CY, Hsieh PO, Chiou DC. Characteristics of ground surface
excavation. In: Proceedings of the ASCE annual convention; 1977 settlement during excavation. Can Geotech J 1993;30(5):758–67.
[preprint]. [20] Peck RB. Deep excavations and tunneling in soft ground. state-of-
[7] Cording EJ. Evaluation and control of ground movements around the-art report. In: Proceedings of the seventh international conference
tunnels and excavations in soil. In: Proceedings of the XII interna- on soil mech. and found. engineering, state-of-the-art volume; 1969.
tional conference on soil mechanics and foundation engineering; p. 225–90.
1985. p. 106–31. [21] Viggiani Gioacchino, Tamagnini Claudio. Ground movements
[8] Cording EJ, Long JH, Son M, Laefer DF. Modeling and analysis of around excavations in granular soils: a few remarks on the influence
excavation-induced building distortion and damage using a strain- of the constitutive assumptions on FE predictions. Mech Cohes Frict
based damage criterion. In: Proceedings of the international confer- Mater 2000;5:399–423.
ence response of buildings to excavation-induced ground movements, [22] Wong KS, Duncan JM. Hyperbolic stress–strain parameters for
CIRIA; 2000. nonlinear finite element analyses of stresses and movements in soil
[9] Desai CS, Zaman MM, Lightner JG, Siriwardance HJ. Thin-layer mass. Geotechnical engineering research report no. TE/74-3, Univer-
elements for interfaces and joints. Int J Numer Anal Method sity of California, Berkeley; 1974.
Geomech 1984;8(1):19–43. [23] Wong KS, Broms BB. Lateral wall deflections of braced excavations
[10] Hashash YM, Whittle AJ. Ground movement prediction for in clay. J Geotech Eng, ASCE 1989;15(6):853–70.
deep excavations in soft clay. J Geotech Eng, ASCE 1996;122(6): [24] Yang G. Analysis of adjacent ground movements for deep excava-
474–86. tions in urban areas. PhD thesis, Seoul National University; 1996.
[11] Hsieh PO, Ou CY. Shape of ground surface settlement profiles caused [25] Yoo C. Behavior of braced and anchored walls in soils overlying rock.
by excavation. Can Geotech J 1998;35:1004–17. J Geotech Geoenviron Eng, ASCE 2001;27(3):225–33.

Você também pode gostar