Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Prediction and Performance of Deep Excavations

for Courthouse Station, Boston


Andrew J. Whittle, M.ASCE 1; Gonzalo Corral, A.M.ASCE 2; Lucy C. Jen, M.ASCE 3;
and Robert P. Rawnsley, M.ASCE 4
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 12/18/14. Copyright ASCE. For personal use only; all rights reserved.

Abstract: Construction of the Silverline Courthouse Station in South Boston involved 18-m-deep excavations at a site underlain by more
than 24 m of normally and lightly overconsolidated Boston blue clay (BBC). The excavations were supported by 27-m-deep floating
diaphragm wall panels and five levels of preloaded cross-lot bracing. This paper compares the measured performance of the excavation
support system with the Class A finite-element (FE) predictions prepared during the original design phase and with the results of Class C
analyses using information obtained during construction. The numerical analyses used data from a special test program of laboratory and in
situ tests at a nearby site. The analyses represent coupled consolidation within the soil mass and the anisotropic stress-strain-strength proper-
ties of BBC using the MIT-E3 soil model. The Class A analyses generally overestimate the lateral wall deflections and underestimate the
measured strut loads, as preloading was not included in the original FE model. However, they provide remarkably consistent predictions of
the measured soil deformations, including settlements, lateral spreading, and subgrade basal heave. The Class C analyses refine the strati-
graphic section, in situ pore pressures, construction time frame, and strut preloads using the data available at the time of construction but make
nominal changes in soil properties. With these limited changes, the Class C model is able to achieve excellent agreement with the measured
data. DOI: 10.1061/(ASCE)GT.1943-5606.0001246. © 2014 American Society of Civil Engineers.
Author keywords: Braced excavation; Field measurements; Finite-element analyses; Clay behavior; Predictions versus measurements.

Introduction Jen (1998) also reported a series of Class A (i.e., a priori) predic-
tions for the proposed design of cross-lot braced excavations for the
Completion of the I-90 connector tunnels (part of the Central Courthouse Station of the Massachusetts bay transportation authority
Artery, Third Harbor Tunnel; CA/T project; 1995–2003) and South (MBTA) South Piers Transitway (Contract Section CC07; Fig. 1),
Piers Transitway (a bus rapid transit system now known as Phase 2 also located less than 400 m from the SBST site. This project involved
of the Silver Line; 2000–2004) provides transportation links that 18-m-deep, 33-m-wide excavations for the station platforms in an area
have radically transformed the South Boston waterfront area, an where deposits of normally to lightly overconsolidated BBC extend to
area now referred to as the Innovation District (Fig. 1). There a depth of more than 40 m. Conventional calculations showed a basal
has been an extensive program of geotechnical research at the stability factor (safety factor, FS) at the lower limit of accepted values
Massachusetts Institute of Technology (MIT) associated with the (conventionally considered in the range FS ¼ 1.20–1.25). The tem-
design and performance of the excavation support systems for these porary works design for this section (Contract CC07) used a floating
tunnel sections. Whelan (1995) documented the measured perfor- support system with four or five levels of cross-lot bracing without
mance of anchored walls for the open-cut boat sections for the I-90 ground improvement. In contrast, the designs for an adjacent contract
connector tunnels (CA/T Contract 4A; Fig. 1), while Jen (1998) (CC05; Fig. 1) used jet grout columns (up to 32 m deep) to stabilize
carried out extensive finite-element (FE) analyses of the project us- the clays below the base of the excavation (Hurley 2004).
ing an advanced effective stress soil model (MIT-E3; Whittle and Figs. 2(a and b) compare one of the original designs analyzed by
Kavvadas 1994; Whittle et al. 1994) to represent the behavior of the Jen (1998) with the final (as-built) support of excavation. Fig. 2(a)
Boston blue clay (BBC). The model input parameters were selected
is referred to as the original design considered in the Class A analy-
using the results from a program of laboratory and in situ tests per-
ses. Based on these analyses, the design engineer actually specified
formed at the South Boston Special Test (SBST) site (Ladd et al.
a minimum of five levels of support with maximum horizontal and
1999) shown in Fig. 1.
vertical spacing between the struts of 6.7 m and 3.3 m, respectively.
1 The contractor selected the final design system based on indepen-
Professor, Civil and Environmental Engineering Dept., Massachusetts
Institute of Technology, Cambridge, MA 02139 (corresponding author). dent analyses. This is referred to as the final design [Fig. 2(b)] and
E-mail: ajwhittl@mit.edu uses 1.22-m-thick diaphragm wall panels (compared to 0.9 m con-
2
Senior Geotechnical Engineer, ITASCA S.A., Dr. José Luis Aguilar sidered by Jen 1998) with larger struts (higher stiffness and allow-
1178, Providencia, Santiago, Chile. able capacity).
3
Lecturer, Civil and Environmental Engineering Dept., Massachusetts The actual excavation was closely monitored through an exten-
Institute of Technology, Cambridge, MA 02139. sive program of field instrumentation with very close control of
4
Formerly Parsons Brinckerhoff, 75 Arlington St., Boston, MA 02116. the excavation process to minimize risks associated with basal in-
Note. This manuscript was submitted on December 28, 2013; approved
stability. The contract specifications also required the contractor to
on October 20, 2014; published online on December 17, 2014. Discussion
period open until May 17, 2015; separate discussions must be submitted for excavate the final stage in a checkerboard pattern when preparing
individual papers. This paper is part of the Journal of Geotechnical and the installation of the base slab for the permanent works. The tem-
Geoenvironmental Engineering, © ASCE, ISSN 1090-0241/04014123 porary works were completed from May 2000 to July 2001, and the
(13)/$25.00. tunnel was opened in 2004.

© ASCE 04014123-1 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng.


Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 12/18/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Site plan of Courthouse Station, South Piers Transitway Project in South Boston (MBTA Contract E02CN14), CA/T tunnels, and special test
site in South Boston (image courtesy of MBTA)

Distance from Centerline [m] 3


60.0 50.0 40.0 30.0 20.0 10.0 0.0 10.0 20.0 30.0 40.0 50.0 [kN/m ]
t

Granular Fill 18.9

30 Cohesive Fill 30
18.1

(b)
Silty Sand 18.9
Diaphragm Wall

Elevation [m MDC]
Width (m) 1.22
20 Toe Elevation (m) +7.6
20
Elevation [m MDC]

Mmax (MN-m/m)
(a) 3.0
Diaphragm Wall Struts (L = 33m, Spacing 6.5m)
Width (m) 0.9 Elevation Preload** Allowable
Level Section
Toe Elevation (m) +10.4 [m] [KN/m] [kN/m]
10 Mmax (MN-m/m) 1.2 =0.61m 10
1 33.2 41 609
BBC w = 1.6cm 18.1
Struts (L = 33m, Spacing 6.7m) =0.91m
2 28.7 110 810
Elevation Preload Allowable* w = 1.6cm
Level Section =0.91m
[m] [kN/m] [kN/m] 3 26.5 429 1223
1 32.0 -- w = 2.5cm
2 28.0 -- 4 23.2 2-W36x135 501 1328
0 5 19.5 2-W36x230 613 1977 0
3 24.1 =0.76m -- 657
w = 1.9cm **Used in Class C analyses only
4 20.1 --
*AISC 9th Ed. (1989)

Glacial Till 18.9


-10 -10

Fig. 2. Comparison of support systems for Courthouse Station excavations, platform section: (a) original design considered in Class A analyses
(data from Jen 1998); (b) final (as-built) design

Corral (2013) has made extensive use of data from the Tran- project is unique among the case studies in Boston in that detailed
sitway CC07 project to understand the performance of floating Class A predictions were produced during the initial design phase
excavation support systems to validate predictions of numerical of the temporary works and can be compared with results from
analyses and to evaluate new procedures for updating predictions more refined analyses performed after completion of the project
using field measurements obtained during construction. This paper (i.e., Class C predictions according to the terminology of Lambe
focuses on the numerical predictions and measured performance for 1973). The results therefore provide a useful basis for evaluating
the platform section of the Courthouse Station shown in Fig. 2. This the predictive capabilities and limitations of the numerical analyses.

© ASCE 04014123-2 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng.


Site Description and Instrumentation to approximately 10–15 g=L below an elevation of þ6 m. Fig. 3(a)
summarizes the stress history profile of the clay based on results of
The MBTA South Boston Piers Transitway Project (Silverline more than 70 one-dimensional (1D) consolidation tests [including
Phase 2) comprises a 1.6 km-long (1-mi), two-lane subway tunnel incremental oedometer, constant rate of strain consolidation
with three underground stations connecting South Station to the (CRSC), and triaxial feedback-controlled K 0 consolidation]. The
new Federal Courthouse and World Trade Center in South Boston. clay can be subdivided into a lower unit (below an elevation of
The cut-and-cover tunnel and station sections were designed to þ6 m) that is very lightly overconsolidated with OCR ≈ 1.0–1.3
accommodate buses (Silverline service to Logan Airport) with pro- and an upper clay crust with preconsolidation rising to σp0 ≈
visions for future conversion to light rail. The Transitway Project 700 kPa at the top of the unit (OCR ≈ 4.5).
is located within generally level ground that was reclaimed during The BBC is underlain by a glacial till with a thickness vary-
the period prior to 1916. The ground surface is at an elevation of ing locally from 5 to 7 m and soft-weathered to hard-fractured
þ34.1 m (Metropolitan District Commission Base, MDC datum) Cambridge argillite (refer to Fig. 8). The glacial till is generally
with the local groundwater table at an elevation of þ31.1 m (close described as a very dense granular material (N̄ ¼ 104  53 bpf;
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 12/18/14. Copyright ASCE. For personal use only; all rights reserved.

to mean sea level). n ¼ 44) containing boulders and cobbles. Core recovery in the
The typical soil profile typically comprises 10–11 m of granular/ argillite ranged from 46 to 100% with the average rock quality des-
cohesive fill materials and silty sand overlying BBC (Fig. 2). The ignation RQD ¯ ¼ 20%.
project site investigation report (Parsons Brinckerhoff, Final The groundwater table in the fill fluctuates seasonally in the
Design Geotechnical Report, South BostonPiers Transitway elevation range of þ29.9 to þ31.1 m (i.e., slightly below mean sea
Project, MBTA Contract E02CN14, Courthouse Station and Tunnel level). Piezometric measurements in the lower glacial till are 1.8 to
Construction Contracts CC07, CC08, and CC10, unpublished 3.0 m below the groundwater table, implying a small head loss
report) characterized three surficial units: (assumed to be linear) through the BBC layer.
1. Granular fill with a thickness of 1.2–2.9 m comprising clean The contract site investigation program included a set of 48 field
dense sand with N̄ ¼ 38  28 bpf (based on data from n ¼ vane shear tests within the 15 geotechnical borings with a program
13 tests). of UU triaxial shear tests on tube samples (Parsons Brinckerhoff,
2. Soft cohesive (hydraulic) fill and/or very soft to medium stiff Final Design Geotechnical Report, South BostonPiers Transitway
organic silt with occasional layers of gray silty sand. The Project, MBTA Contract E02CN14, Courthouse Station and Tunnel
combined thickness ranges from 4 to 7.6 m with N̄ ¼ Construction Contracts CC07, CC08, and CC10, unpublished re-
4  4 bpf (n ¼ 49). port). These data can be compared with data from a much more
3. Medium dense, silty sand with a thickness ranging from 0.9 to extensive research program at the SBST site (Haley and Aldrich,
6.4 m with N̄ ¼ 29  13 bpf (n ¼ 40). “Final geotechnical engineering report, central artery (I-93) / tunnel
The main unit of BBC ranges in thickness from 22.9 to 37.1 m. (I-90) project, design sections D001A and 0001 B, Boston, Mas-
The physical and index properties of this marine illitic clay have sachusetts,” submitted, Massachusetts Highway Department; Ladd
been extensively documented elsewhere. Ladd et al. (1999) report et al. 1999). The latter includes tests on samples obtained from
a plasticity index I P ranging from 20 to 35% with a specific gravity fixed piston and block samplers and is the first program to compare
Gs of 2.79  0.01 and a 50% clay fraction. The liquidity index properties of K 0 -consolidated specimens prepared using SHANSEP
(I L ) generally falls between 0.3 and 0.8, increasing with depth. and recompression procedures. The results from triaxial and
The pore fluid salt concentration decreases from approximately direct simple shear devices also enable direct assessment of the
30 g=L (compared to 35 g=L for seawater) at the top of the clay undrained strength anisotropy for BBC. Fig. 3(b) summarizes the

In Situ Stresses and Pre-Consolidation (kPa) Undrained Shear Strength, s (kPa) In-situ Hydraulic Conductivity, k (m/sec)
u
v0
-10 -9 -8
0 200 400 600 0 50 100 150 200 10 10 10
0
FE Models Lab Data: Special Test Program - Lab Tests Laboratory Hydraulic Conductivity
Class A SB Special Test Program (Ladd et al. 1999) Line Type Site Source
' (Ladd et al., 1999) Triaxial Compression, suTC Direct Simple Shear, suDSS
30
5 v0 Class C CRS CA/T C04A Whelan (1995)
Oedometer SHANSEP:
SHANSEP: CRS I-95 Morrison (1984)
CRSC S=0.28, m=0.68 S=0.20, m =0.77 (Crust)
S=0.18, m=0.66 (Lower) Const. H I-95
u0 CK0UC (triaxial) Recompression CRS I-95 Varney (1998) 25
10
Piezometer Data
(March - July 1999)
suDSS
Project Elevation (m)
20
15 suTE
Depth (m)

15
Elevation (m)

20 suTC

10
25
CC07 Site Investigation
'p ± 1SD (PBQD, 1998)
Field Vane, s
uFV 5
30 TW-19/20
TW-12
TW-2/5/21
MIT-E3 0
35 (Jen, 1998)

Class A&C
-5
40 FE Models

(a) (b) (c)

Fig. 3. In situ stresses and properties of BBC: (a) stress history; (b) undrained shear strength; (c) hydraulic conductivity

© ASCE 04014123-3 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng.


undrained shear strength profiles from SHANSEP parameters for
undrained triaxial compression (CK0 UC) and direct simple shear
(CK0 UDSS) modes. For a given mode of shearing, there is a small
decrease in su through the clay crust to a local minimum at an
elevation of þ7 m (suDSS ¼ 50 kPa). The undrained strength in tri-
axial compression is typically 50–60% higher than that measured in
direct simple shear (suTC ¼ 120 kPa versus suDSS ¼ 75 kPa at the
base of the clay—i.e., suDSS =suTC ¼ 0.625). The results of recom-
pression tests are consistent with SHANSEP strength profiles
(for the triaxial compression mode) but generally show more scatter
associated with sample disturbance. The field vane data are in rea-
sonable agreement with the SHANSEP DSS parameters in the clay
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 12/18/14. Copyright ASCE. For personal use only; all rights reserved.

crust but are highly scattered in the lower BBC (results affected by
stress changes in the more sensitive lower BBC) and are much less
reliable for modeling purposes.
There were no laboratory or in situ hydraulic conductivity mea-
surements for BBC in the site investigations for the CC07 project.
The in situ vertical hydraulic conductivity, kv0 , was estimated from
laboratory CRSC tests at the SBST site (Haley and Aldrich, “Final
Geotechnical Engineering Report, central artery (I-93)/tunnel
(I-90) project, design sections D001A and 0001 B, Boston,
Massachusetts,” submitted, Massachusetts Highway Department;
Whelan 1995) by extrapolating measured e- log k relations to the
in situ void ratio e0 . Fig. 3(c) shows the values of kv0 in the range
of 3 × 10−8 to 3 × 10−7 cm=s and a trend of a decrease of the hy-
draulic conductivity with the depth.
Fig. 4(a) shows the instrumentation location plan in the central
area of the Courthouse Station, referred to as the platform section.
There are a series of low-rise buildings located 50–60 m from the
south wall and one small structure (Seaman’s Chapel) to the north.
These existing buildings are brick construction with heights of up
to 6 stories and are founded on spread footings. The instrumenta-
tion used to monitor the excavation performance for the platform
section included:
1. Lateral movements of the diaphragm wall by five inclinometers
cast within the wall; three in the north wall (I-05N1, I-06N1,
and I-07N1); and two in the south wall (I-06S1 and I-07S1).
2. Lateral deformations in the soil by two inclinometers (I-06N2
and I-06S2) located approximately 40 m behind each wall (all (a)
inclinometers extend into the underlying glacial till).
3. Surface settlements by surveys of 16 deflection monitoring
points (DMPs) and by a series of survey nails located along
the tops of the diaphragm wall panels (DMP1).
4. Excavation heave by one multipoint heave gauge (HV-6C)
located at the center of the excavation.
5. Piezometric head measurements from one vibrating wire
piezometer inside the excavation (PZ-06 C).
6. Strut forces measured by 20 vibrating wire strain gauges dis-
tributed at each level of bracing and at every second strut line
in the platform section. The final design uses single-pipe struts
for the first three levels of bracing and pairs of wide-flange
beams at Levels 4 and 5 [Fig. 2(b)].
The aerial photo in Fig. 4(b) shows the temporary works after
excavation to final grade elevation and gives a clear perspective on
the adjacent structures and site conditions.

FE Models
The main objective of this paper is to evaluate the capabilities of (b)
FE models for predicting the performance of the excavation support
Fig. 4. (a) Instrumentation plan for platform area of Courthouse
system at the platform section, comparing predictions obtained
Station; (b) aerial view of Courthouse Station excavations, September
prior to construction (Class A) versus a posteriori calculations
2001 (image courtesy of JF White Contracting)
(Class C). In general, constitutive modeling of the soil behavior and

© ASCE 04014123-4 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng.


the selection of model input parameters represent a major source of Fig. 2(a) shows the reference support system for the original design
uncertainty affecting FE predictions. This project is unusual in this with the soil profile, referred to as Case A1. The analyses assume
respect, given the proximity of the Courthouse Station to the SBST plane-strain conditions with horizontal soil horizons and symmet-
site (Fig. 1), the wealth of in situ and laboratory test data reported ric loading conditions. The calculations were performed using the
in Fig. 3, and the extensive prior research in the development and ABAQUS program (Hibbitt, Karlsson and Sorensen 1998) with
application of advanced effective stress soil models that have been quadrilateral isoparametric finite elements (8-4 node mixed ele-
calibrated for BBC. Whittle et al. (1993) and Hashash and Whittle ments for the soils and 8-node solid elements for the elastic dia-
(1996) also reported results of analyses illustrating the factors phragm wall). The diaphragm walls panels are wished-in-place,
controlling the performance of floating excavation support systems and there is no slip between the concrete panels and adjacent soils
in BBC. (rough interface). Table 1 provides the elastic properties of the
All of the analyses reported in this paper (both Classes A and C) struts. The glacial till serves as a rigid base with a constant piezo-
use the MIT-E3 soil model (Whittle and Kavvadas 1994; Whittle metric head H ¼ 30.5 m. The simulated construction sequence as-
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 12/18/14. Copyright ASCE. For personal use only; all rights reserved.

et al. 1994) to describe the behavior of BBC, while simpler/more sumes an overdig of 0.6 m prior to installation of each level of the
conventional elastic, perfectly plastic models are used for the upper struts, zero precompression of the struts, and a fixed time frame of
soil units, where there is a scarcity of engineering property data. 45 days for each excavation level, with an additional 90 days at the
The predictions simulate the coupled flow and deformation to final grade elevation (i.e., the total construction time frame is
represent partial drainage effects within the clay. The Class A pre- 315 days from the start of excavation for the original design com-
dictions are based exclusively on information available at the time pared to 360 days for the final design).
of design; they assume a specific time sequence for the excavation Table 2 summarizes the key input parameters and models used
process and adopt the conservative assumption that there is no pre- to describe the soil behavior in the Class A predictions. The upper
loading of the cross-lot bracing. In contrast, the Class C analyses soil units (fill layers and silty sand) are described using a simple
include refinements in the stratigraphic model (reflecting data from linear elastic–perfectly plastic (EPP) soil model using correlations
installation of monitoring instruments), include measured preloads with the SPT data (Stroud 1989) and prior empirical experience
at each strut level, and simulate the actual timeline of construction from projects in the Boston area (Einstein et al. 1983; Whittle et al.
events (excavation and support). The following paragraphs detail 1993). The deformation and shear strength properties of the upper
differences in the Class A and Class C models. soil units affect the performance of the shallow excavations (less
than 10 m deep), and material properties were selected to achieve
reasonable agreement with the measured wall deflections in the
Class A Predictions prior projects. The current study focuses on the behavior for a much
Jen (1998) carried out a series of analyses to examine the impact deeper excavation (with a floating support system) that is controlled
of changes in the diaphragm wall embedment depth, vertical strut by the properties of the underlying marine clay. BBC is modeled
spacing, uncertainties in the undrained shear strength, and effects of using MIT-E3 (Whittle and Kavvadas 1994), a generalized effec-
partial drainage on the design of the excavation support system. tive stress soil model that describes the rate-independent behavior

Table 1. Stiffness Properties of Struts


Strut Average spacing, Cross-sectional Axial stiffness,
Case level Section s (m) area, A (cm2 ) EA=s (MN=m)
A1 1–4 ϕ ¼ 0.76 m; w ¼ 1.9 cm 6.7 445 1,330
A2 and C 1 ϕ ¼ 0.61 m; w ¼ 1.6 cm 6.45 296 920
2 ϕ ¼ 0.91 m; w ¼ 1.6 cm 448 1,400
3 ϕ ¼ 0.91 m; w ¼ 2.5 cm 709 2,200
4 2-W36 × 135 512 1,600
5 2-W36 × 230 872 2,700
Note: Young’s modulus of steel E ¼ 200 GPa; pipe struts have diameter ϕ and wall thickness w.

Table 2. Soil Models and Input Parameters for Class A Analyses


Unit Top elevation(m) Model γ t (kN=m3 ) c 0 (kPa) [su =σv0
0 ] 0 (°)
ϕPS OCR K0 0
G=σv0 ν0 k (m=day)
Granular fill 34.1 EPP 18.9 0 30 — 0.50 35 0.3 0.3
Cohesive fill 32.3 EPP 18.1 [0.22] — — 0.75 20 0.3 0.009
0 15° 0.3
Silty sand 25.6 EPP 18.9 0 35 — 0.50 35 0.3 0.3
BBC 21.9 MIT-E3 18.1 Appendix — 3.00 0.84 Appendix 4.3 × 10−5
19.5 [Fig. 3(b)] 2.52 0.77
17.1 2.14 0.71
14.6 1.79 0.66
12.2 1.46 0.61
9.8 1.25 0.57
7.3 1.0 0.53
Glacial till −7.7 EPP 18.9 0 43 — 1.0 250 0.3 0.09
Argillite — Elastic 21.8 — — — 1.0 10,000 0.3 0.009
Note: EPP = elastic–perfectly plastic model (Drucker-Prager yield in ABAQUS (Hibbitt, Karlsson and Sorensen 1998); Mohr-Coulomb yield in plaxis);
boldface text = alternate properties used in Class C analyses; concrete diaphragm walls E ¼ 2.26 × 104 MPa; ν 0 ¼ 0.15.

© ASCE 04014123-5 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng.


2.0
Measured Data; BBC using the PLAXIS FE program (Brinkgreve et al. 2008). This pro-
Sym. Shear Mode gram uses 15-3 mixed interpolation triangular elements to model
Plane Strain Compression
Triaxial Compression
the coupled flow and deformation within the soil mass and 15-node
Direct Simple Shear solid elements (i.e., cubic strain interpolation) for the nonporous di-
Plane Strain Extension aphragm wall. The standard version of the linear EPP soil model in
Undrained Strength Ratio, su/ 'vc

1.5
Triaxial Extension
PLAXIS uses a conventional Mohr-Coulomb shear strength envelope
MIT-E3 Predictions TC
(referred to as the MC model) with zero dilation, while earlier analy-
PSC ses with ABAQUS (Hibbitt, Karlsson and Sorensen 1998) use a
Drucker-Prager strength criterion. However, the two models are
1.0 DSS matched to achieve the same shear strength under plane-strain
conditions. Apart from this difference in detail, the two Class A
analyses—referred to as Cases A1 and A2—differ only in the speci-
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 12/18/14. Copyright ASCE. For personal use only; all rights reserved.

fication of the excavation support system [Figs. 2(a and b)].


Figs. 6(a and b) summarize the predictions of the lateral wall
0.5 deformations, bending moments, and strut loads for Class A analy-
ses of the original and final designs of the excavation support
PSE system (Cases A1 versus A2). The final design has a deeper toe
embedment for the diaphragm wall (elevations of þ7.6 m versus
TE
þ10.4 m) and a thicker wall section (1.22 m versus 0.9 m) and
0.0
1 2 3 4 5 6 7 8 uses five levels of struts (with a higher axial stiffness). These mod-
Overconsolidation Ratio, OCR ifications are effective in reducing the computing wall deflections
by 16–18 mm [Fig. 6(a)] at later stages of the excavation but have
Fig. 5. Comparison of MIT-E3 predicted and measured undrained little influence on the mode of deformation of the diaphragm wall.
shear strength ratios for resedimented BBC (adapted from Whittle et al. The final design generates larger bending moments associated with
1994, © ASCE) higher bending stiffness and greater embedment of the diaphragm
wall panels [Fig. 6(b)]. Analysis for the original design shows that
the allowable load is exceeded at the fourth level of struts [Fig. 2(a)];
of normally and moderately overconsolidated clays. The model de- hence, there is clear logic for recommending an extra level of strut-
scribes a number of important aspects of soil behavior that have ting, as used in the final design. It can also be seen that the fifth strut
been observed in laboratory tests, including small strain nonlinear- level carries the highest design load (938 kN=m; Fig. 6).
ity and anisotropic stress-strain-strength properties associated with Once the excavation reaches the penultimate stage (grade at
a 1D consolidation history. Whittle et al. (1994) describe the selec- an approximate elevation of þ20 m), there are similar magnitudes
tion of model input parameters for BBC based on laboratory tests of deflections at the grade elevation and at the toe of the wall.
on resedimented clay specimens and show extensive comparisons Excavation to the final grade (elevation of þ16.5 m) involves large
between computed and measured effective stress paths and stress- inward movements at the toe. These results highlight the impor-
strain behavior in a variety of shear modes. They report that the tance of basal stability in the design of the floating support system.
MIT-E3 model gives excellent predictions of the measured un- Table 3 summarizes the results of stability calculations for the origi-
drained stress-strain-strength properties for K 0 normally consoli- nal and final support systems using the average undrained shear
dated BBC, but predictions deviate from the measured shear strength profile, suDSS [Fig. 3(b)]. The calculations were performed
strengths at higher OCRs. Fig. 5 compares the computed undrained using numerical limit analyses (Ukritchon et al. 2003) and account
shear strength ratios (su =σvc 0 ) versus OCR for BBC in a range
for the maximum allowable bending capacity of the diaphragm wall
of shear modes. The results show that MIT-E3 overestimates the panels (Fig. 2). The results show mobilization factors, FS (partial
undrained strength ratio at OCR ¼ 4.0 by 15–25% compared to factors of safety on the undrained shear strength) to ensure stabil-
measured data from compression (plane strain and triaxial) and ity of the support system. Lower-bound and upper-bound analyses
simple shear (DSS) modes. In applying the model for analyses of show that the original design has FS ¼ 1.21–1.37, while the final
the Transitway excavations, Jen (1998) adjusted the stress history design produces a modest increase in the mobilization factor,
state variables for the model (i.e., OCR and K 0 ) to achieve good FS ¼ 1.26–1.42. Figs. 7(a and b) show that the critical failure
agreement with the undrained shear strength profiles presented in mechanisms are very similar in both cases, involving the formation
Fig. 3(b). The appendix summarizes the input material constants of a plastic hinge in the wall at the lowest strut line, with failure
used for the MIT-E3 model (unchanged from Whittle et al. 1994), extending to the base of the clay.
while Table 2 provides the initial values of the stress history state The effects of undrained strength anisotropy on stability have
variables assumed in the current calculations. The analyses effec- been considered based on procedures recommended by Ukritchon
tively assume lower values of σp0 through the clay crust compared et al. (2003). Numerical limit analyses were carried out using the
to data shown in Fig. 3(a) to achieve more reliable modeling of Davis-Christian criterion (Davis and Christian 1971) to represent
undrained properties in the clay crust. Table 2 shows that the an- variations in the mobilized undrained shear strength with the direc-
alyses also assume a constant hydraulic conductivity (k ¼ 4.3 × tion of shearing based on behavior computed by the MIT-E3 soil
10−8 m=day) through the full depth of the BBC unit. This assump- model (with the OCR profile in Table 2 and a specified shear strain
tion is consistent with measurements by Whittle et al. (2001) at the γ of 1%). Table 3 shows that the resulting mobilization factors are
I-95 test site but is at the lower end of the range reported by Whelan very similar to values obtained from isotropic analyses using suDSS .
(1995) [Fig. 3(c)].
In actuality, the design of the excavation support was amended
prior to construction, as shown in Fig. 2(b). Corral (2013) carried Class C Analyses
out Class A analyses for the final design section using the same The Class C analyses (Corral 2013) incorporate refinements in the
modeling assumptions as Jen (1998). These analyses were performed model stratigraphy based on inclinometer installation records. Fig. 8

© ASCE 04014123-6 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng.


0

30
5
Depth Below Ground Surface [m]

25
10

Elevation [m MDC]
20
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 12/18/14. Copyright ASCE. For personal use only; all rights reserved.

15

15
20

10
25 (a) (b)

100 80 60 40 20 0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0


Wall Deflection, (mm) Bending Moment, M (MN-m/m)
w

Class A Predictions Class A Predictions


'Original Design': A1 Final Design: A2 Max Strut Load (KN/m)
(Jen, 1998) (Corral, 2013) Strut Level Case A1 Case A2
Exc. (m) Line Exc. (m) Line 1 284 172
32.0 32.0 2 448 214
27.5 29.0 3 701 346
23.5 25.9 4 735 434
19.5 22.9 5 N/A 938
16.5 19.8
16.5

Fig. 6. Summary of Class A predictions (no preloading of struts)

shows that the surface of the glacial till dips 5 m from the south to of strutting [Fig. 2(b)]. However, the model does not simulate de-
north wall—a feature that affects the symmetry assumed in prior tails of the checkerboard excavation (and base slab installation)—
analyses. The Class C FE model is extended into the upper weath- an approximation that affects the results computed for the last
ered rock to allow for drained relaxation of stresses from the prior excavation stage. The Class C analyses assume that the upper soil
assumption of K 0 conditions. The model also includes surcharge units (granular and cohesive fills and silty sand) are fully drained
loads (q ¼ 25 kPa acting at the top of the silty sand layer) to layers. The MC model is used to represent the stress-strain pro-
represent stresses induced by the 6-story buildings on the south perties of the glacial till and rock layers with input parameters
side of the excavation (Fig. 4). Piezometer monitoring data (from estimated from prior empirical experience (Einstein et al. 1983)
the period prior to construction) show lower piezometric heads (Table 2). There are no changes in the MIT-E3 parameters
in the underlying glacial till (H ¼ 28.2 m). These are assumed as for BBC.
boundary conditions at the base of the model, with a linear head
drop through the clay layer.
The Class C analyses simulate the interpreted time frames of Comparison of Measurements and Predictions
construction (Fig. 8) and average values of preloads at each level
Figs. 9(a–f) compare the predicted and measured horizontal deflec-
tions at the north and south walls of the platform section at the last
Table 3. Mobilization Factors on Undrained Shear Strength of Soil for three stages of excavation (Stages 4–6). Corral (2013) provides full
Stability of Support Systems details of the earlier stages of excavation that are affected by the
Design Original Final/amended Final/amended properties of the upper soil units (fill and silty sand), which were
simulated using a simple EPP model with empirical parameters
Mobilized
based on performance data from prior shallow excavation proj-
Clay shear Isotropic: Isotropic: anisotropica
strength suDSS suDSS at γ ¼ 1.0%
ects. The figures combine data from three inclinometers installed
through the north diaphragm wall panels into the underlying argil-
Lower bound 1.211 1.259 1.241 lite (I-05N1, I-06N1, and I-07N1; Fig. 4) with two in the south wall
Upper bound 1.368 1.415 1.395 (I-06S1 and I-07S1). The figures compare the data with predictions
a
Based on MIT-E3 predictions using OCR profile in Table 2. from Class A (Case A2) and Class C analyses.

© ASCE 04014123-7 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng.


The Class A predictions generally overestimate the inward de-
formations of the walls above the excavated grade but are in good
agreement with the measured toe movements through Stage 5.
The analyses overestimate incremental toe movements in the
Stage 6 excavations. The Class C analyses are in excellent
agreement with the measured wall deflections at all stages of ex-
cavation. The differences between the computed Class A and C
wall deflections are due, in large part, to modeling of the strut
preloads. The Class C analysis also more accurately represents
the horizontal displacements in the underlying soils due to refine-
ments in modeling the stratigraphy of the basal layers (glacial till
and rock) but still tends to overestimate the deformations
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 12/18/14. Copyright ASCE. For personal use only; all rights reserved.

measured in the lower BBC.


Lateral deformations in the retained soil were only measured
intermittently over the course of the project. Fig. 10 summarizes the
computed horizontal soil deformations at locations approximately
40 m from the north and south walls with the measured deformations
at the final excavated grade (Stage 6). The data show deformations
increasing monotonically from the base of the clay with maximum
movements, δ h ≈ 26 mm, at the ground surface. The results are in
excellent agreement with the Class A predictions, while the Class C
predictions tend to underestimate movements within the upper soil
units. This effect may be attributable to the assumption of drained
behavior in the cohesive fill in the Class C model.
Surface settlements were measured by a series of DMPs
installed within the pavements and sidewalks surrounding the site
[Figs. 4(a and b)] and at locations along the tops of the diaphragm
wall panels. Figs. 11(a–f) compare the computed and measured set-
tlements at each of the last three stages of excavation. There is sig-
Fig. 7. Upper-bound stability analyses for original and final designs nificant scatter in the measured data among points equidistant from
from numerical limit analyses the diaphragm walls, particularly on the south side, where DMPs
are located within the sidewalks adjacent to buildings. Settlements
are also generally higher on the south side than the north side, and
The inclinometer data show very consistent results along the this may reflect effects such as surcharge loads from the buildings
platform section. Maximum wall deflections—δ w ¼ 50–52 mm (included in Class C analyses) or differences in the groundwater
(for four of the five inclinometers)—occur at the toe of the dia- table across the site (i.e., interference in groundwater seepage con-
phragm wall panels at the last stage of excavation. There is little ditions due to tunnel construction).
difference in the measured performance of the south and north The Class A analyses are in remarkably good agreement with
walls (but slightly more scatter in the data for the latter). The data the measured data at final grade elevation. However, these results
interpretation assumes that there is zero displacement at the rock-till are closely correlated with inward movements of the diaphragm
interface (with less than 6 mm measured at the top of the glacial till). wall (the top of the diaphragm wall panels are predicted to settle

Fig. 8. Cross section for Class C analyses

© ASCE 04014123-8 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng.


Lateral Wall/Soil Movement (mm)
North South
-20 0 20 40 60 80 80 60 40 20 0 0
0

10
10

20
20

(a) (b)
Measured Data Measured Data 30
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 12/18/14. Copyright ASCE. For personal use only; all rights reserved.

30 I06N1 I06S1
I05N1 I07S1
I07N1 Predictions
Predictions Class C
Class C Class A 40
40 Class A

0 0

10 10

Depth (m)
Depth (m)

20 20

30 30

40 40
(c) (d)
0 0

10 10

20 20

30 30

40 40
(e) (f)
-20 0 20 40 60 80 80 60 40 20 0 -20
Lateral Wall/Soil Movement (mm)

Fig. 9. Comparison of computed and measured lateral wall deflections at last three stages of excavation: (a and b) Stage 4 excavation CD 139-153;
(c and d) Stage 5 excavation CD 183-201; (e and f) Stage 6 excavation CD 274-286

by 10 mm; Figs. 11(e and f). The Class C analyses estimate smaller [Figs. 11(c–f)] but also tend to overestimate heave movements at
maximum surface settlements on the north side than on the south Stage 4 [Figs. 11(a and b)].
side [15 mm versus 20 mm; Figs. 11(e and f), respectively]. While Vertical soil movements beneath the base of the excavation
the Class C analyses do not match the outer envelope of the were monitored at four elevations within the BBC using a multi-
measured data, they appear to provide very reasonable predictions point heave gauge [HV-6C, with a datum at the till-rock interface;
of the overall settlement troughs. The Class C analyses predict Fig. 4(a)], while pore pressures were obtained nearby (PZ-06C)
very small net vertical movements (<3 mm) at the tops of the dia- using vibrating wire piezometers at two elevations. Fig. 12 com-
phragm walls, consistent with the measured data at Stages 5 and 6 pares the predicted and measured subgrade heave during the last

© ASCE 04014123-9 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng.


Lateral Soil Movement (mm)
0 20 40 60 80 80 60 40 20 0 0
0

10 10

Depth (m)

Depth (m)
20 20

(a) (b)
30 Measured Data Measured Data 30
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 12/18/14. Copyright ASCE. For personal use only; all rights reserved.

I06N2 I06S2
(37m from wall) (41m from wall)
Predictions Predictions
Class C Class C
40 Class A Class A 40

Fig. 10. Comparison of computed and measured lateral soil deformations at locations 40 m from the diaphragm walls

Horizontal Distance from Wall (m)


-140 -120 -100 -80 -60 -40 -20 0 0 20 40 60 80 100 120 140
10 10
North South

0 0

-10 -10

-20 (a) (b) -20


DMP Data DMP Data
Predictions Predictions
-30 Class A Class A -30
Class C Class C

-40 -40
10 10
North South

Surface Settlement (mm)


Surface Settlement (mm)

0 0

-10 -10

-20 -20

-30 -30

(c) (d)
-40 -40
10 10
North South

0 0

-10 -10

-20 -20

-30 -30

(e) (f)
-40 -40
-140 -120 -100 -80 -60 -40 -20 0 0 20 40 60 80 100 120 140
Horizontal Distance from Wall (m)

Fig. 11. Comparison of predicted and measured surface settlements at excavation for Stages 4–6

© ASCE 04014123-10 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng.


Excavation Heave (mm) Date
-20 0 20 40 60 80 100 120 Nov/1/2000 Jan/5/2001 Mar/11/2001 May/16/2001 Jul/20/2001 Sep/24/2001
40
0
240
Measured Data - PZ06C
30 Elevation +13.7m
200
Predictions
Class A
10 Stage 4 30 160
CD 139-153 Class C
Stage 5
CD 183-201 Stage 6
20 120

Project Elevation (m)


CD 274-286
Depth (m)

80
20 20
10 40

Pore Pressure (kPa)


Piezometric Head (m)
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 12/18/14. Copyright ASCE. For personal use only; all rights reserved.

0
30 (a)
10
Measured Data 40
0
HV-06C
Predictions Measured Data - PZ06C
Elevation + 7.6m 280
Class A
40 Predictions
Class C
Class A 240
-10 30
Class C
200
Fig. 12. Comparison of predicted and measured below-grade heave at
center of excavation for Stages 4–6 160

20
120

80
three stages of excavation. The measured data show heave in the
upper BBC crust exceeding 70 mm at the final stage of excavation. (b) 40
The Class C predictions are in excellent agreement with 10
0 50 100 150 200 250 300 350
the measured data at all three stages. The Class A analyses
Time (days)
are also in reasonable agreement with the measured data
through the fifth stage but overestimate ground movements Fig. 13. Comparison of predicted and measured pore water pressure-
in the final stage of excavation. Refinements in the model time response in clay beneath center of excavation
stratigraphy explain the improved agreement achieved in the
Class C predictions.
Figs. 13(a and b) compare the computed and measured time
histories of pore pressures for piezometers at elevations of 2000
þ13.7 m and þ7.1 m, respectively. The measured data show Measured (±1SD)
net decrements of pore pressures Δu=σv0 ≈ 0.5 (at both eleva- Allowable Load
tions) due to the excavation process. The Class A analyses over-
Class A
estimate the initial pore pressures at both depths and clearly
Maximum Strut Load (kN/m)

1500
Class C
provide only a crude estimated timeline of events based on design
assumptions. However, pore pressure changes at each phase of
excavation are well-estimated through Stage 5. The analyses
overestimate pore pressure changes at Stage 6 but reliably show 1000
the rebound in pore pressures due to subsequent partial drainage
within the clay (there is very little recovery of pore pressure at
an elevation of þ7.6 m, close to the center of the clay layer)
[Fig. 13(b)]. The Class C predictions match the initial pore 500
pressures and accurately describe the pore pressure-time history
at both depths. Small discrepancies at Stage 6 reflect details of the
controlled excavation sequence to the final grade that were not
present in the FE model. 0
Strut loads are interpreted by averaging the measured data from Strut Level 1 Strut Level 2 Strut Level 3 Strut Level 4 Strut Level 5
20 vibrating wire strain gauges installed at different strut lines.
Fig. 14. Comparison of computed and measured maximum strut loads
The measured data correspond to total loads due to thermal ex-
at each level of strutting (note: allowable loads include both design
pansion of the struts and lateral earth pressures. Following the
earth pressures and thermal stresses)
methodology proposed by Hashash et al. (2003), it is estimated
that thermal loads range from 11% of the maximum loads in Strut
Level 1 decreasing to 4% at Level 5. These effects are relatively
small and have little influence on the interpretation of the numeri- Levels 2–4) while overestimating wall deflections. The predicted
cal analyses. The measured data show significantly more variabil- toe movements during Stage 6 excavation generate large strut
ity at Strut Levels 4 and 5 than at Strut Levels 1–3, Fig. 14. The loads at Level 5, and these are in surprisingly good agreement
Class A analyses ignore preloading of the struts and hence gen- with the measured data. The Class C analyses use the average
erally underestimate the maximum strut forces (readily apparent at measured preloads [Fig. 2(b)] as initial conditions at the time of

© ASCE 04014123-11 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng.


strut installation. Predictive capabilities can then be assessed by presented in this paper include refinements in the stratigraphy
subsequent changes in strut loads and the estimated maximum and in situ pore pressures (based on data from installing monitoring
strut forces. The Class C analyses are in very good agreement instruments) with interpreted construction events (time frames
with the maximum strut loads measured at all levels but notably of excavation steps and strut preloads) but make only nominal
overestimate measurements at Level 5 (by 20–25%). This result changes to the modeling of the soil properties. The key findings
may also be attributed to the checkerboard sequencing used can be summarized as follows:
during the final stage of excavation but not considered in the 1. The Class A predictions, using an advanced effective stress
two-dimensional (2D) FE models. soil model that is well-calibrated using high-quantity test data,
can achieve realistic and consistent predictions of performance
for a braced excavation system as interpreted from measure-
Conclusions ments of wall deflections, ground movements (surface settle-
ments and horizontal movements in the retained soil), and pore
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 12/18/14. Copyright ASCE. For personal use only; all rights reserved.

There is a scarcity of published case studies comparing the mea- pressures. The strut loads can also be well-predicted, assuming
sured performance of well-instrumented excavation support sys- that there is advanced knowledge of preload forces at the time
tems with numerical predictions. This paper summarizes a detailed of design.
set of comparisons that have been achieved for one critical section 2. The Class A predictions performed previously by Jen (1998),
of the MBTA South Piers Transitway Project (platform area of the which considered several parametric numerical analyses, were
Courthouse Station), which relied on a floating support system extremely useful and relevant for the final geotechnical design.
and careful construction and monitoring procedures to mitigate In fact, they enabled construction of the project without resort-
potential problems associated with a basal stability safety factor ing to expensive ground improvement techniques.
at the lower limit of acceptable limits (during the final stage of 3. The Class C analyses achieve a significant improvement in
excavation). the prediction of wall deflections and strut loads and modest
The research on this project extends back over a period of improvements in the ground movements (surface settlements
more than 15 years. The project provided a unique opportunity to and subgrade heave).
evaluate Class A predictions using an advanced effective stress 4. The Class C analyses described in this paper could have
soil model (MIT-E3), with input parameters obtained from a very been performed at the time of construction, as they involve no
extensive program of laboratory and in situ tests on the behavior iterations on uncertain soil properties for the deep layer of mar-
of natural BBC (Ladd et al. 1999). The Class A predictions (after ine clay (although the properties of the upper fill and sand
Jen 1998) involved a series of careful engineering judgments to layers were based on prior empirical experience for shallow
ensure realistic representation of the undrained shear properties of excavations). In this respect, they offer conclusive evidence
the 30-m-deep clay layer. The analyses included careful studies on of the predictive accuracy of the proposed finite-element
the effects of partial drainage (over the expected time frame of models.
construction), strut spacing, and toe embedment. Corral (2013) The results of this study provide strong evidence to support the
updated the Class A predictions for the as-built structural support integrated use of numerical analyses and monitoring data with con-
system (referred to as the amended design) and carried out detailed trolled construction procedures for reducing the risks associated
Class C predictions for the project. The Class C predictions with marginal geotechnical stability conditions.

Appendix. MIT-E3 Input Parameters for BBC (Data from Whittle et al. 1994)

Parameter/ Boston blue


Test type symbol Physical contribution/meaning clay (BBC)
One-dimensional consolidation eo Void ratio at reference stress on virgin 0.988
(oedometer, CRSC, and so forth) consolidation line
λ Compressibility of virgin, normally consolidated clay 0.184
C Nonlinear volumetric swelling behavior 22.0
n 1.6
h Irrecoverable plastic strain 0.2
K 0 oedometer or K 0 triaxial K 0NC K 0 for virgin, normally consolidated clay 0.53
2G=K Ratio of elastic shear to bulk modulus (Poisson’s ratio 1.05
for initial unload)
0
Undrained triaxial shear tests: ϕTC Critical state friction angles in triaxial compression 33.4°
OCR ¼ 1, CKoUC; OCR ¼ 1, 0
ϕTE and extension (large strain failure criterion) 45.9°
CKoUE; OCR ¼ 2, CKoUC c Undrained shear strength (geometry of bounding 0.866
surface)
St Amount of postpeak strain softening in undrained 4.5
triaxial compression
ω Nonlinearity at small strains in undrained shear 0.07
γ Shear-induced pore pressure for OC clay 0.5
Shear wave velocity κ0 Small strain compressibility at load reversal 0.001
Drained triaxial ψ0 Rate of evolution of anisotropy (rotation of bounding 100
surface)
Note: Input parameters based on laboratory tests on resedimented BBC.

© ASCE 04014123-12 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng.


Acknowledgments Hashash, Y. M. A., and Whittle, A. J. (1996). “Ground movement predic-
tion for deep excavations in soft clay.” J. Geotech. Eng., 10.1061/
The authors are grateful to the MBTA for supporting the original (ASCE)0733-9410(1996)122:6(474), 474–486.
Class A analyses for Section CC07 of the South Piers Transitway Hibbitt, Karlsson and Sorensen. (1998). ABAQUS version 5.8 user’s
Project. Parsons Brinckerhoff were the design engineers and were manual, Providence, RI.
responsible for monitoring the performance of the excavations. The Hurley, T. M. (2004). “Jet grout bottom seal for cut-and-cover tunnel.”
Proc., North American Tunneling, L. Ozdemir, ed., 265–270.
construction was carried out by JF White Contracting Company.
Jen, L. C. (1998). “The design and performance of deep excavations
The views and opinions expressed in this paper are those of the in clay.” Ph.D. thesis, Dept. of Civil and Environmental Engineering,
authors and are not those of the organizations or other individuals Massachusetts Institute of Technology (MIT), Cambridge, MA.
responsible for the completion of this project. Mary Ainsley Ladd, C. C., Young, G. A., Kraemer, S. R., and Burke, D. M. (1999).
(MBTA), Richard O’Brien (PB), and Youssef Hashash (now at “Engineering properties of Boston blue clay from special testing
UIUC) were instrumental in setting up the original study. The sec- program.” Proc., Special Geotechnical Testing: Central Artery/Tunnel
ond author (G. C.) was partially supported by a scholarship through Project in Boston, Massachusetts, ASCE, Reston, VA, 1–24.
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 12/18/14. Copyright ASCE. For personal use only; all rights reserved.

the Beca Presidente de la República de Chile for his Ph.D. studies at Lambe, T. W. (1973). “Predictions in soil engineering.” Geotechnique,
MIT. The authors would also like to thank Boonchai Ukritchon and 23(2), 149–202.
Yixing Yuan, who performed the basal stability calculations, and Morrison, M. J. (1984). “In situ measurements on a model pile in
clay.” Ph.D. thesis, Massachusetts Institute of Technology (MIT),
Sherif Akl and Paul Bonnier, who integrated the MIT-E3 model in
Cambridge, MA.
the PLAXIS code. Stroud, M. A. (1989). The standard penetration test—Its applications and
interpretations, Thomas Telford, London, 29–49.
Ukritchon, B., Whittle, A. J., and Sloan, S. W. (2003). “Undrained stability
References of braced excavations in clay.” J. Geotech. Geoenviron. Eng., 10.1061/
(ASCE)1090-0241(2003)129:8(738), 738–755.
AISC (American Institute of Steel Construction). (1989). Specification for Varney, A. (1998). “A performance comparison between a novel tapered
structural steel building: Allowable stress design and plastic design, piezoprobe and the piezocone in Boston blue clay.” SM thesis, Dept.
Chicago. of Civil and Environmental Engineering, Massachusetts Institute of
Brinkgreve, R. B. J., Broere, W., and Waterman, D. (2008). PLAXIS: Finite Technology (MIT), Cambridge, MA.
element code for soil and rock analyses, version 9.0, Balkema, Whelan, M. P. (1995). “Performance of deep excavations in Boston.” SM
Rotterdam, Netherlands. thesis, Dept. of Civil and Environmental Engineering, Massachusetts
Corral, G. (2013). “Methodology for updating numerical predictions of Institute of Technology (MIT), Cambridge, MA.
excavation performance.” Sc.D. thesis, Dept. of Civil and Environ- Whittle, A. J., DeGroot, D. J., Ladd, C. C., and Seah, T.-H. (1994). “Model
mental Engineering, Massachusetts Institute of Technology (MIT), prediction of the anisotropic behavior of Boston blue clay.” J. Geotech.
Cambridge, MA. Eng., 10.1061/(ASCE)0733-9410(1994)120:1(199), 199–224.
Davis, E. H., and Christian, J. T. (1971). “Bearing capacity of anisotropic Whittle, A. J., Hashash, Y. M. A., and Whitman, R. V. (1993). “Analysis of
cohesive soil,” J. Soil Mech. Found. Div., 97(5), 753–769. deep excavation in Boston.” J. Geotech. Eng., 10.1061/(ASCE)0733-
Einstein, H. H., Azzouz, A. S., McKown, A. F., and Thompson, D. E. 9410(1993)119:1(69), 69–90.
(1983). “Evaluation of design and performance—Porter square transit Whittle, A. J., and Kavvadas, M. J. (1994). “Formulation of MIT-E3 con-
station chamber lining.” Proc., Rapid Excavation Tunneling Conf., stitutive model for overconsolidated clays.” J. Geotech. Eng., 10.1061/
Vol. 1, 597–620. (ASCE)0733-9410(1994)120:1(173), 173–198.
Hashash, Y. M. A., et al. (2003). “Temperature correction and strut loads Whittle, A. J., Sutabutr, T., Germaine, J. T., and Varney, A. (2001).
in central artery excavations.” J. Geotech. Geoenviron. Eng., 10.1061/ “Prediction and interpretation of pore pressure dissipation for a tapered
(ASCE)1090-0241(2003)129:6(495), 495–505. piezoprobe.” Geotechnique, 51(7), 601–617.

© ASCE 04014123-13 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng.

You might also like