Geotechnical Engineering Proceedings of the Institution of Civil Engineers

Volume 168 Issue GE1 Geotechnical Engineering 168 February 2015 Issue GE1
Pages 3141 http://dx.doi.org/10.1680/geng.13.00045
Seismic stability of braced excavations next Paper 1300045
to tall buildings Received 09/04/2013 Accepted 11/04/2014
Published online 15/08/2014
ORiordan and Almufti Keywords: dynamics/retaining walls/seismic engineering

ICE Publishing: All rights reserved

Seismic stability of braced

excavations next to tall
buildings
Nick ORiordan PhD, CEng, FICE, PE(Calif.) Ibrahim Almufti MS, LEED AP, PE, SE(Calif.)
Director, Ove Arup and Partners, London, UK Senior Engineer, Arup North America Ltd, San Francisco, CA, USA

It can take many years to complete the construction of deep, wide and long excavations in urban environments. In
seismically active areas, there is a finite probability that a significant earthquake could take place during the
construction period. The designer and builder are confronted with the need not only to maintain control of ground
and retaining wall movements during the normal construction conditions but also to protect surrounding buildings
during a seismic event. It follows that the seismic loads from adjacent structures and their foundations must be
considered together with the available load pathways through the propping system. The paper considers the
available methods of design and proposes a performance-based design philosophy in which the temporary shoring
system and permanent installation are designed for static loading conditions and for the seismic condition only in as
much as the performance objectives of adjacent buildings are achieved for various levels of earthquake shaking. An
example of the use of performance-based design is presented for the effects of a 130 m tall building of 30 m square
plan upon the temporary shoring system for a long and wide excavation. The results obtained are compared with
code-based calculations and general conclusions are drawn.

Notation the excavation is open, as well as during the operation of the

B width of excavation completed underground structure. Design codes do not provide
G secant shear modulus of soil at current shear strain g adequate coverage for either case.
Gmax maximum shear modulus, obtained from shear wave
test measurements This paper provides approaches for the seismic design of multi-
g gravitational acceleration propped, deep excavations adjacent to tall buildings in urban
N selected return period used in Figure 4 areas. It describes the selection of ground motion and the
n number of years excavation is open assessment of the systems response under a seismic event. Figure
Sa spectral acceleration 1 illustrates common soilstructure interaction situations requir-
T fundamental/first mode period of structure ing analysis.
wmax maximum soil displacement caused by excavation
Adjacent buildings
annual rate of occurrence of earthquake
v90 initial vertical effective stress
max maximum shear stress
Ground
1. Introduction Basement Tunnel Pile displacement
Limited available vacant space in urban areas compels many vital
facilities, such as public transit systems, parking and building
facilities, to be constructed underground. The construction of
such facilities typically requires deep, wide and long excavations
that can take many years to complete. Large excavations in an
urban setting where the excavation site is surrounded by many
buildings are very challenging. Excavations that are not ade-
quately shored may adversely affect the surrounding buildings Closed arrows indicate Open arrows indicate
(Ciria, 2001; Davis and Henkel, 1982; Finno et al., 2010). The free-field displacement inertial force
effects
challenge is further complicated when the effect of seismic
interaction between excavations and adjacent buildings is taken Figure 1. Seismic ground displacements imposed on below-
into account. In seismically active areas there is a finite ground structures (modified after Free et al., 2001)
probability that a significant earthquake could take place while

31
Downloaded by [] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 Geotechnical Engineering Seismic stability of braced excavations
Volume 168 Issue GE1 next to tall buildings
ORiordan and Almufti

2. Current methodologies for retaining B /2 B

structures in far-field conditions Adjacent building
Conventionally, retaining structures are designed using Mononobe and foundations
Okabi (MO) (Seed and Whitman, 1970) or Wood (1973) equations A B
depending, respectively, on whether the excavated structure is
considered to be flexible, mobilising active pressures, or is
sufficiently stiff to behave rigidly under seismic loading. Sitar et al.
(2012) have challenged these design methods using careful model-
D Excavation
ling in the centrifuge. They conclude C
wmax F depth

(a) the MO method, simplified according to Seed and Whitman wmax

(1970), is appropriate for retaining structures of depth, H, up wmax
to 7 m, with the resultant force applied H/3 from the base H
(b) the Wood (1973) solution is not representative of conditions
commonly encountered in practice and its continued use is E
not recommended
Figure 2. Illustration of the potential participation of adjacent
(c) caution is needed when applying the MO method to retained
building in the incremental displacement field for a wide
depths greater than 7 m, recognising that at some depth the
excavation (based on Bolton et al., 2008)
structure will move in a complex manner with the soil.

To deal with retaining structures of greater than 7 m depth, site

response analyses are used with selected ground motions to ment (w) is partially driven by the foundation of an adjacent
predict soil deflection profiles in the undeveloped, green or structure if it falls anywhere within the zone bounded by the
far-field situation. The far-field displacements are then applied surface ACEDB. The magnitude of the displacement caused by
as prescribed displacements in a pseudo-static, finite-element, the adjacent structure will be governed by its location within the
continuum model. The stiffening (or softening effect) of the zone, its stiffness and weight. Ukritchon et al. (2003) examined
retaining structure is then examined, to establish a system of the stability of a number of multi-propped excavations and found
design forces acting upon the buried structure (Anderson et al., that most operated under a two-dimensional factor of safety of
2008; Free et al., 2001). around 1 .1 to 1 .3 on operational strength of soil in the zone
bounded by EFIH. The distance AB in Figure 2 can be con-
p
Callisto and Soccodato (2010) describe a methodology to convert sidered to be B/ 2 where B is not less than the half-width of the
pseudo-static displacements obtained by these methods into a excavation. The building weight can be represented as a uni-
design. Unfortunately, again as illustrated in Figure 1, the formly distributed vertical stress added to the mechanism
proximity of large, often very tall buildings is such that in no way bounded by ACEDB. It follows that, if that vertical stress exceeds
can the conditions of the far field pertain, and superposition of the operational strength of the soil mobilised by that mechanism,
displacement fields will underestimate the interaction effect. then the static stability may be compromised and the seismic
loading will further increase the transient and permanent defor-
Hashash et al. (2010) have extended Anderson et al. (2008) to mation of the wall (represented in part by the dashed line between
include various configurations of buried structure, including the D and E in Figure 2). In turn, this will increase the prop loads
increased racking effect from the inclusion of embedded shoring and wall demands between props.
walls. They also compared pseudo-static behaviour with that
predicted by dynamic analysis in plane strain situations. However, Where the seismic loads from a building are derived from codes
the effect of adjacent structures is not considered. and then applied to the design of an adjacent excavation (Figure
2), the effects of soilstructure interaction must be considered. In
3. Static stability and pseudo-static analysis general, buildings designed using codes rely on estimates of
of excavated structures adjacent to accelerations derived for free-field surface motions. This neglects
buildings the kinematic effects of embedded basements, which may signifi-
Analysis of seismic stability can be complex. In many cases a cantly reduce the demands.
simple, pseudo-static treatment can be appropriate if there is only
limited participation of the adjacent structure in the excavation Stewart et al. (1998) examined the performance of individual
process under static conditions. Figure 2 shows the displacement buildings with embedded basements to strong motions in Califor-
field of Bolton et al. (2008), an extension of the various base nia. This work is being extended by the activities of ATC-83
stability mechanisms considered by Ukritchon et al. (2003). The (NIST, 2012); however, the interaction of adjacent buildings or
distribution of prop forces and wall demand is a function of the underground installations remains at the level of research. Stewart
soil and wall displacement that follows excavation. The displace- et al. (1998) considered standard code calculation procedures in

32
Downloaded by [] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 Geotechnical Engineering Seismic stability of braced excavations
Volume 168 Issue GE1 next to tall buildings
ORiordan and Almufti

which the foundation is assumed to be rigid, fixed at the base and structures with foundations on linear elastic foundation soils (e.g.
subject to free-field representation of ground acceleration. They Yahyai et al., 2008). These tend to overstate the amount of
found that interaction at in-plane separation distances less than the width of
the smaller structure. Trombetta et al. (2012) carried out
(a) the fundamental period of the structure increases with the centrifuge testing of rigid dissimilar structures founded on pads
explicit inclusion of a foundation system and at higher aspect on dense sand with a site period of 0 .6 s. They modelled a low-
ratios (height/width), decreasing the flexural demands in the rise structure adjacent to a 29 .6 m high structure at prototype
building scale. These tests showed that, where the PGA exceeded 0 .3g,
(b) there was little apparent difference between structures on key parameters such as structural drift, foundation rotation and
piled and shallow foundations. structural demand are adversely affected.

In addition, higher mode response, which generally governs the Considerable effort is required to generalise these overall trends
shear demands in tall buildings, may be greatly reduced when in behaviour for above-ground structures. Further complications
kinematic effects are accounted for deep embedded basements arise where the adjacent structure involves the removal of soil
(Stewart and Tileylioglu, 2007). and an unloading stress path in the adjacent founding soil. The
magnitude of the SSSI effect can dominate the design of the
This means that the use of code-based seismic forces in a two- excavated structure.
dimensional pseudo-static analysis will tend to produce an over-
estimate of the demand on the structural support system for the The design of the excavation support system and the associated
underground structure where permanent underground installation adjacent to structures that
have a period, T, of more than 2 s is most economically carried
j structural foundations fall within the zone ACEDB of Figure 2 out using SSSI techniques. The process comprises the following
j the vertical stress that it imposes is lower than the operational steps
strength of the soil.
(a) selection of ground motions and return period
For situations with an operational soil strength of about 100 kPa (b) establishment of design basis and properties of adjacent
below line CD in Figure 2, adjacent buildings lower than about structures, including foundations
15 storeys would fall into this category. (c) computer simulation of excavation process and earthquake
performance
Pseudo-static analysis can be appropriate for buildings with (d ) comparison of predicted earthquake performance of adjacent
fundamental period, T , 2 s, provided that the inclusion of the structure with design basis
building load does not reduce the factor of safety against basal (e) completion of design of temporary support and permanent
heave to below 1 .3. Furthermore, owing to the potential for installation.
conservatism in static design procedures, Bray et al. (2010) have
suggested that seismic earth pressures on retaining walls can be Figure 3 shows the code-intended seismic performance objectives
ignored where the peak ground acceleration (PGA) is less than for new buildings in the USA. These are designed under the implicit
0 .3g. It is therefore unlikely that seismic interaction with such objective of life safety performance in a design earthquake.
comparatively modest adjacent buildings will be a governing load Although it is not explicitly related to a specific intensity level, the
case.
Performance level
For taller and/or heavier structures, the pseudo-static load effect Ground Immediate Collapse
Operational Life safety
motion occupancy prevention
can be examined by choosing a reasonable load pathway for the
seismic loads and by assuming that the load is transmitted
Frequent
uniformly through the soil through a 40458 wedge. For a raft OCII: Ordinary
foundation at ground level, the addition of base shear is OCIV: Essential
OCIII: High occupancy
straightforward. Ways of including lateral loading effect from
adjacent piled buildings under static loading conditions are Design
discussed by Kaul (2010) for metro stations.

4. Seismic performance criteria MCE

Adjacent buildings apply base overturning moments and shear
loads that will usually be out of phase with each other and with
the response of the excavated structure. These effects are often Figure 3. Matrix for code-intended performance objectives
referred to as structuresoilstructure interactions (SSSIs). Recent (FEMA, 2009)
publications on numerical analysis of SSSIs have been limited to

33
Downloaded by [] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 Geotechnical Engineering Seismic stability of braced excavations
Volume 168 Issue GE1 next to tall buildings
ORiordan and Almufti

design earthquake is approximately equivalent to a 475-year return used for the temporary condition and ASCE 37-02 (ASCE, 2002)
period earthquake (10% probability of exceedance in 50 years) in implies 20%. To put these recommendations into the context of
California. For comparison, Eurocode 8 (BSI, 2004) adopts a return probability of exposure, Figure 4 examines the situation for the
period of 475 years explicitly and similarly intends for life safety urban centre of San Francisco, whereby acceleration spectra are
performance. Building performance under frequent earthquakes derived for return periods of 50 and 100 years, normalised with
is rarely considered, although the implicit performance expectation the spectrum for a return period of 475 years. It can be seen that
is that new buildings would suffer minimal damage (ASCE 7-10 the Eurocode 8 and ASCE 37-02 recommendations, shown
(ASCE, 2010)) and Eurocode 8 has a damage limitation require- bracketed in Figure 4, are reasonable for situations where the
ment associated with a return period of 95 years. fundamental period of the system is greater than 2 s, but not for
systems with shorter fundamental periods. These include deep
For building structures, a performance-based design is used as a excavations and situations in which adjacent building perform-
means to demonstrate that the performance objectives intended ance is dominated by higher modes of vibration. For these, it may
by prescriptive code-based design are explicitly met. The per- be more appropriate to define a uniform hazard spectrum for the
formance objectives are defined by the amount of tolerable desired intensity level, as follows.
damage a building sustains for a particular severity of shaking.
To demonstrate the performance objectives are met, a non-linear The frequent-level earthquake ground motion is not defined in the
response history analysis (NLRHA) is typically required. building codes, but several performance-based design guidelines
(CTBUH, 2014; PEER TBI, 2014) recommend a return period
An analogous methodology may be employed for open excava- between 43 and 72 years. Table 1 gives the probability that the
tions in urban areas and enables a focused dialogue to take place
07
between the owner of the underground structure and the adjacent
building owner. The general performance-based philosophy is to
Ratio N-year/475-year RP spectral acceleration

demonstrate that the seismic performance of surrounding build- 06

ings only is not compromised by the excavation and is no worse
relative to the performance of the building in the existing
05
condition (i.e. no excavation or shoring system), under a pre-
100/475
specified severity of ground motion (e.g. frequent level). The
props and shoring system need not achieve any independent 04
performance objectives separate from the adjacent buildings
performance objectives (i.e. in an urban area, this implies that
03
stability of the excavated system must not be compromised). This 50/475
design intent may be attained by the following measures.
02
(a) Design the props and shoring system for the gravity forces Sa are average horizontal
associated with the open excavation in the static condition, spectra, GMRot150, using
01 NGA attentuation
ensuring that deflections are kept to acceptable limits. relationships for
(b) Check that the adjacent building sustains no more damage San Francisco downtown
under the temporary level earthquake event (see below) in 0
the temporary condition compared to the existing condition 0 2 4 6 8 10 12
Period: s
(no excavation). Since it is likely that the structural details
(and thus non-linear properties of the structural components) Figure 4. Variation of spectral acceleration ratio with period
in the adjacent building are not available (especially for older
existing buildings), it may be necessary to gauge the
performance of the building by ensuring the seismic-induced
lateral and vertical settlements are within some acceptable Number of years excavation Probability of exceedance
limit to be agreed upon by the owner/engineer. is open while excavation is open: %
(c) Check that the seismic-induced demands on the props and
1 2
shoring system in the temporary condition are less than
2 4
their capacities.
3 6
(d ) The process should be repeated for each ground motion under
4 8
consideration, but no less than three motions should be used.
5 10
4.1 Ground motions Table 1. Probability of exceedance of earthquake ground motion
Eurocode 8 (BS EN 1998 (BSI, 2004)) recommends 30% of the while excavation is open assuming a 50-year return period
design actions associated with a return period of 475 years are

34
Downloaded by [] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 Geotechnical Engineering Seismic stability of braced excavations
Volume 168 Issue GE1 next to tall buildings
ORiordan and Almufti

frequent-level, 50-year earthquake ground motion will be exceeded 5.1 Effects of the excavation on soil response
during the period that the excavation remains open. Table 1 shows The response for the soil column directly below the excavation in
that the hazard level for the excavation depends on the length of the long-term future loading condition may not be the same as
time the excavation remains open, as well as on the earthquake predicted by the free-field motions at the state of stress in the
intensity level. The hazard level is not defined solely by adopting a ground pre-excavation. The excavation has the effect of reducing
pre-determined earthquake intensity. For comparison, the 475-year the effective stresses in the soils to a level below which they
return period earthquake has a 10% chance of being exceeded in cause the stiffness and strength to degrade.
the typically stated lifetime of a building (50 years). Table 1 shows
that the utilisation of a 50-year return period earthquake to analyse A one-dimensional site response analysis may also be performed
an excavation that is only open for a few years may be conservative using a reduced height of the soil column based on the completed
compared to traditional risk tolerance. The values in Table 1 were excavation depth to assess the potential of liquefiable soils within
determined using the following equation the excavation and at depth. The soil properties should be adjusted
to account for the reduction in stress and changes to other
PE 1e(n) engineering properties. The shear modulus reduction curve should
be adjusted so that the implied shear strength or apparent friction
angle is consistent with the new engineering properties and stress
where PE is probability of exceedance, is annual rate of state for each layer. The change in the small strain shear modulus
occurrence and n is the number of years the excavation is open. can be estimated using many of the available correlations from
The return period is calculated by 1/. Seed and Idriss (1970) to Darandeli (2001). An example of the
variation in shear modulus with increasing shear strain for a clay
A hazard level that aligns more closely with the traditional risk of medium plasticity and moderate over-consolidation ratio is
tolerance may be obtained as follows. given in Figure 5, in which the curve for static analysis (simulat-
ing the excavation sequence) is compared to that used for subse-
(a) Use a temporary level earthquake ground motion to assess quent dynamic analysis at any stage of excavation.
the seismic response of the adjacent building during the
temporary excavation. Where the results of one-dimensional analysis show that a layer
(b) Define this temporary level ground motion as having a 10% is liquefiable, its shear strength should be neglected when
probability of exceedance while the excavation is open. excavation stability is considered. A one-dimensional site re-
(c) Adopt a lower damping (e.g. 2% rather than the traditional sponse analysis excludes the stiffness of the shoring and bracing
5%) when generating the response spectrum of the adjacent system and should always be complemented by a more complete,
building. The building is not expected to yield during a low- three-dimensional soilstructure interaction analysis as follows.
intensity earthquake and so significant hysteretic damping is
unlikely to occur. 6. Structuresoilstructure interaction
(d ) An SSSI analysis may then be carried out as described below, analysis
to determine the structural demands on the walls and props. Consider, as an extreme example, the tall, heavy building,
These should then be compared to the yield capacities of the approximately 35 m 3 60 m in plan, located 10 m outside an
individual members and their connections. 18 .3 m deep excavation illustrated in Figure 6. In this situation,

5. Free-field site response 10

One-dimensional wave propagation analyses are a standard tool
to understand the effects of soil stiffness and thickness on the
08
seismic energy at a location of interest. The analyses can be
performed using equivalent linear (Idriss and Sun, 1992) or non-
06
G/Gmax

linear (Kwok et al., 2006) approaches using conventional wave

propagation softwares, such as Shake 2000, Deepsoil, Siren and
04
LS-Dyna. Understanding the assumptions and simplifications
used in each approach is important when reviewing the results. Static shear stiffness
02
Dynamic shear stiffness
These analyses assume that there is no excavation, no shoring
system and there are no adjacent structures. The results of the 0
106 105 104 103 102 101
analyses are the free-field seismic response, which can be used as Shear strain,
input in the soilstructure interaction analysis. The results of the
analyses can be checked by comparing the surface response Figure 5. Examples of non-linear secant shear modulus
spectra with spectra established through conventional code-based degradation in static and dynamic loading for stiff clay
procedures.

35
Downloaded by [] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 Geotechnical Engineering Seismic stability of braced excavations
Volume 168 Issue GE1 next to tall buildings
ORiordan and Almufti

Adjacent structure

Sand 1
Clay 1
Sand 2
Clay 2
Sand 3
Sand 4
Potential
failure surface

Clay 3 Buried shear

walls

Clay 4

Bedrock

Figure 6. Tall building of equivalent vertical static load of

500 kPa adjacent to 20 m deep excavation, showing extent of
buried shear walls to prevent basal heave

the building load applies a vertical static load of 500 kPa. This is
approximately four times the operational shear strength of the
soils at excavation level, which would be intersected by the
potential failure surface shown in Figure 6. In the static
condition, the stability of the excavation is slightly less than 1,
and additional deep support is needed. The first mode period of
the building is approximately 4 .5 s and the far-field site period is
approximately 1 .5 s.

6.1 Buried shear walls

The soil strength under the excavation, in which zone the
potential failure surfaces EFIH in Figure 2 would otherwise form,
is enhanced using shear walls, approximately 2 m wide at 4 .5 m
centres with average concrete strength properties of 28 MPa
embedded inside the retaining wall to a depth of 60 m below the
excavation, into the underlying bedrock to prevent failure in the
static condition. A three-dimensional dynamic analysis using LS-
Figure 7. Computer simulation model (LS-Dyna)
Dyna (Figure 7) is carried out with unidirectional ground motion
perpendicular to the excavation to examine the permanent
displacements that are induced by the earthquake and conse- j pressure-sensitive non-linear shear stress/strain soil properties
quently any additional demands on the props and shoring system. j degrading G/Gmax curves where Gmax is established from
For this study, the ground motions associated with a 100-year shear wave velocity tests
return period were used because they had already been developed j pore pressures and excess pore pressures associated with the
for the site for separate purposes. Since the excavation is shear strain rates pertaining during excavation and seismic
expected to be open for approximately 5 years, this corresponds excitation stages
to a ground motion with 5% probability of exceedance. j foundation of surrounding building, including appropriate
cracked concrete properties and piles
The assessment of excavation behaviour under seismic loading is j representation of the adjacent building, including equivalent
best completed using NLRHA. The software, as a minimum, mass and stiffness to capture the dynamic behaviour
explicitly includes j prop and shoring wall system

36
Downloaded by [] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 Geotechnical Engineering Seismic stability of braced excavations
Volume 168 Issue GE1 next to tall buildings
ORiordan and Almufti

31 500 Temporary condition

j interface layers between structural elements and virgin soils
that represent disturbed soil Existing condition no excavation
27500
j soil domain sufficiently large that boundary effects are
negligible under seismic loading. 13 500

Base shear: kN
The model for the adjacent building in this analysis may often be 4500
0
based on elastic properties because non-linearity in the structural
4 500
components is usually minor under frequent level shaking.
Damping is assumed to be 2% of critical which is consistent with 13 500
new research on tall buildings.
22 500
Details of the NLRHA are given in the appendix to this paper.
31 500
The results show that the overall performance of the building was 0 10 20 30 40 50
better in the temporary condition with the open excavation than it Time: s
was in the existing condition with no excavation. The improve-
ment results from the addition of the buried shear walls to the Figure 9. Shear at base of tower (Kobe earthquake input): elastic
shoring wall system (see Figure 6). During the earthquake, plastic base shear force at the underside of the mat of the adjacent
strains develop in the highly stressed soils beneath the building building
(Figure 6) and result in permanent settlements. Figure 8 shows
the settlements from one of the ground motion scenarios. The
vertical, seismically induced settlement with the temporary buried shear walls and in the original condition without an
excavation and shear walls present is slightly less than it would excavation. The maximum seismically induced base shear force
have been if there were no excavation. The other two ground in the direction of the excavation is about 27 500 kN, or an
motion scenarios exhibited similar response. average of 785 kN/m run of the 35 m wide piled mat.

Figure 9 shows the elastic base shear forces at the underside of Next, the seismic demands in the props are investigated to ensure
the adjacent tall building mat arising from earthquake loading. that they are within capacity. Figure 8 shows the prop locations
Base shear is similar in both the temporary situation with the and Table 2 provides a summary of predicted loads at completion

Sand 40 m
1 Prop 1 35 m Basement level (B1)
Prop 2
40 m 0
Seismically induced settlement below mat: mm
Clay Prop 3
1 40 m
Prop 4
Sands 45 m
2 and 3 25

Clay
2 50

75
Clay
3
58 m

Imperial Valley 100-year 100

Soil replaced with concrete Temporary condition
57% treatment 216 m wide
125
Existing condition
Clay no excavation
4 Soil replaced with concrete 150
86% treatment 55 m wide

10 20 30 40 50 60 70 80 90 100
Bed
rock Distance from excavation: m

Figure 8. Seismic-induced settlements in temporary and existing

conditions

37
Downloaded by [] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 Geotechnical Engineering Seismic stability of braced excavations
Volume 168 Issue GE1 next to tall buildings
ORiordan and Almufti

Prop Depth: m Completion of excavation (LS-Dyna): Seismic load increment (LS-Dyna): Seismic load increment
kN/m run kN/m run (Wood, 1973):a kN/m run

Remote from Adjacent to Remote from Adjacent to Remote from

building building building building building

1 4 .0 160 90 80 10 550

2 7 .5 410 390 310 40 410
3 11 .5 870 560 220 160 270
4 15 .5 660 750 30 560 330
a
Using the pressure distribution suggested by Matthewson et al. (1980).

Table 2. Summary of predicted prop loads at completion of

excavation and after seismic shaking

of excavation and the increment added to those loads after compress the bottom prop. The rotational flexibility of this
seismic shaking. Values are expressed in kN/m run of wall and building may be greater than most, because the piles are not
are given for two locations: adjacent to the building and remote deeper than the zone CDE illustrated in Figure 2.
from it, where the buried shear walls are absent. The total force
in the remote props is the greater. This is due to the local This combination of static and dynamic analyses has enabled the
reduction of lateral stress from the building basement excavation examination of the permanent displacements and associated seis-
and the mobilisation of lateral resistance through the piled mat of mic force distributions from a tall, heavy building onto an adjacent
the building and the buried shear wall system. multi-propped excavation. The following conclusions are drawn.

Figure 10 shows how the prop forces change during an earth- (a) The seismic forces taken by the lowest level of props amount
quake. The lowest level of props, which are approximately 2 .4 m to about 75% of the NLRHA base shear force under the
on centre, are resisting a significant proportion of the seismic adjacent building mat.
load while the loads in the higher levels of props reduce during (b) The seismic force attracted to the buried shear wall system
and after seismic shaking. This behaviour, together with the force below the excavation that is, the difference between the
distributions remote from the building, are insightful results from NLRHA calculated base shear force and the force taken by
NLRHA and are not as expected from code-based formulations. the props is approximately 25% of the base shear force.
These results can be attributed to three-dimensional effects and, (c) The horizontal line of reaction to the base shear is close to
after seismic shaking, to the rotational flexibility of the adjacent the base of the piles of the adjacent building, and this does
building at its base, which tends to unload the top props and not conform to the force distribution predicted using
conventional pseudo-static formulations such as Mononobe
0 Okabe (Seed and Whitman, 1970) or Wood (1973).
(d ) The seismic effects on the stability of a multi-propped
excavation from adjacent, tall, heavy buildings can be
900
Prop force: kN/prop

controlled by the construction of buried shear walls within the

shoring wall system, and this has the potential to improve the
Prop No. 1 performance of the adjacent buildings under seismic loading.
1800
Prop No. 2
Prop No. 3 7. Conclusion
2700 Prop No. 4 Most underground structures built in urban environments will
require a temporary shored excavation due to spatial restrictions.
The shoring system is designed to provide global stability and to
3600 meet performance criteria for static conditions. Meeting the
0 10 20 30 40 50 60 performance criteria can be critical to the success of an excava-
Time: s
tion, especially when existing structures are present within the
Figure 10. Development of seismic-induced forces in props at zone of deformations outside the excavation.
2 .4 m centres (note: static excavation-induced prop forces are
shown at time 0) It is suggested that pseudo-static analysis can be appropriate for
buildings with T , 2 s, provided that the inclusion of the building

38
Downloaded by [] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 Geotechnical Engineering Seismic stability of braced excavations
Volume 168 Issue GE1 next to tall buildings
ORiordan and Almufti

05 Free field
10 One-dimensional site response analysis
04
09
Acceleration: g

08

Spectral acceleration: g
03
07
02 06
05
01
04
0 03
001 01 1 10 02
Period: s
01
Figure 11. 100-year site-adjusted top of rock mean equal hazard 0
response spectrum 001 01 1 10
Period: s

Figure 12. Surface spectra: comparison between free-field

load does not reduce the factor of safety against basal heave to response in LS-Dyna and representative one-dimensional site
below 1 .3. For taller, heavier structures, detailed, dynamic, three- response analysis
dimensional analysis is required.

Numerical analysis tools now allow SSSI analyses that incorpo- Acknowledgements
rate dynamic, seismic shaking. These tools can help in under- The authors are grateful to their colleagues Michael Willford,
standing the complex interaction of the soil, shoring system for Brian Simpson, Payman Khalili-Tehrani, Yuli Huang, Kirk Ellison
the excavation, the completed underground structure and adjacent and Stephen McLandrich in supporting this research. The authors
buildings. The shoring system for a new excavation adjacent to also thank the Transbay Joint Powers Authority for the opportu-
tall, heavy buildings with T . 2 s and insufficiently deep founda- nity to publish some of the findings arising from the design of the
tions in areas with PGA . 0 .3g will require an augmented Transbay Transit Center, San Francisco in this paper.
shoring system to take account of seismic interaction effects.
After the underground structure has been constructed, the effects Appendix: Details of non-linear response
of new, tall buildings to be sited adjacent to that structure can be history analysis used in the example
evaluated. Tall, heavy buildings with T . 2 s in areas with (a) Bedrock input motion (see Figure 11).
PGA . 0 .3g may well require foundations that are significantly (b) Free-field response in LS-Dyna compared to representative
deeper and stiffer than would be required to satisfy only static one-dimensional site response analysis (Deepsoil) (see
loading constraints. Figure 12).

PGA: g (Kobe (Nishi-Akashi)) max / vo : g (Kobe (Nishi-Akashi)) Max shear strain: % (Kobe (Nishi-Akashi))
0 005 010 015 020 025 030 0 005 010 015 020 025 0 02 04 06
0 0 0 0
Fill 125
20
Bay
mud
45
50 Marine 50 50
sands
Lower 65
Bay mud 75
Lower
marine sands
92
100 100 100
Depth: ft

Depth: ft
Depth: ft

Deep soil Old Deep soil Deep soil

bay
Siren clay Siren Siren
unit I
150 150 150
170
Old
bay
200 clay 200 200
unit II

Weathered 233
bedrock
250 250 250 250

Figure 13. Site response analysis

39
Downloaded by [] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 Geotechnical Engineering Seismic stability of braced excavations
Volume 168 Issue GE1 next to tall buildings
ORiordan and Almufti

06
Free field Under tower
embedded retaining walls subjected to seismic loading.
Proceedings of Workshop on Eurocode 8: Perspectives from
05
an Italian Standpoint, Naples, Italy, pp. 110.
Spectral acceleration: g

Ciria (2001) Building Response to Tunnelling: Case Studies from

04
Construction of the Jubilee Line Extension, London (Burland
JB, Standing JR and Jardine FM (eds)). Thomas Telford,
03
London, UK.
Darandeli MB (2001) Development of a New Family of
02 Normalized Modulus Reduction and Material Damping
Curves. PhD thesis, University of Texas, Austin, USA.
01 Davis RV and Henkel DJ (1982) Geotechnical problems associated
with the construction of Chater Station, Hong Kong. The
0 Arup Journal 17(1): 410.
001 01 1 10
Period: s FEMA (Federal Emergency Management Agency) (2009) NEHRP
Recommended Seismic Provisions for New Buildings and
Figure 14. Response spectrum at bottom of mat elevation Other Structures. FEMA, Washington, DC, USA, FEMA
P-750 Report.
Finno RJ, Hashash YMA, Arduino P and Bellevue WA (eds)
(c) Site response analysis (one-dimensional analyses shown: (2010) Proceedings of the 3rd Earth Retention Conference.
results are similar in LS-Dyna) (see Figure 13). ASCE, Reston, VA, USA.
(d ) Response spectrum generated from LS-Dyna analysis at the Free MW, Pappin JW, Sze JWC and McGowan MJ (2001) Seismic
bottom of the mat (underside of basement B1 on Figure 8) design methodology for buried structures. Proceedings of the
(see Figure 14). 14th Southeast Asian Geotechnical Conference, Hong Kong,
pp. 509514.
Note that the response under the building is lower than in the free Hashash YMA, Karina K, Demetrious K and ORiordan N (2010)
field owing to the local influence of the mass of the building, the Seismic design considerations for underground box
stiffness of the piled foundation and the buried shear walls within structures. Proceedings of the ASCE Earth Retention
the adjacent excavation. Conference, ER2010, Seattle (Finno RJ, Hashash YMA,
Arduino P and Bellevue WA (eds)). ASCE, Reston, VA,
REFERENCES USA, pp. 620637.
Anderson DG, Martin GR, Lam I and Wang JN (2008) Seismic Idriss IM and Sun JI (1992) SHAKE91: A Computer Program for
Analysis and Design of Retaining Walls, Buried Structures, Conducting Equivalent Linear Seismic Response Analyses of
Slopes and Embankments. National Cooperation Highway Horizontally Layered Soil Deposits. Department of Civil and
Research Programme (NCHRP), Transportation Research Environmental Engineering, University of California, Davis,
Board, Washington, DC, USA, Report 611. CA, USA.
ASCE (American Society of Civil Engineers) (2002) ASCE 37-02: Kaul K (2010) Cut and Cover Metro Structures: Geo-Structural
Design loads on structures during construction. ASCE, Design: An Integrated Approach. Spon, London, UK.
Reston, VA, USA. Kwok OA, Stewart JP, Hashash MA et al. (2006) Practical
ASCE (2010) ASCE/SEI 7-10: Minimum design loads for implementation of analysis routines for nonlinear seismic
buildings and other structures. ASCE, Reston, VA, USA. ground response analysis. Proceedings of the 8th National
BSI (2004) BS EN 1998: Eurocode 8: Design of structures for Conference on Earthquake Engineering, San Francisco.
earthquake resistance. BSI, London, UK. Earthquake Engineering Research Institute, Oakland, CA,
Bolton MD, Lam SY and Osman AS (2008) Supporting USA, Paper S46.
excavations in clay from analysis to decision making. Matthewson MB, Wood JH and Berrill JB (1980) Seismic design
Proceedings of the 6th International Symposium on of bridges earth retaining structures. Bulletin of the New
Geotechnical Aspects of Underground Construction in Soft Zealand Society of Earthquake Engineering 13(3): 280293.
Ground, Shanghai, vol. 1, pp. 1225. NIST (2012) Soil-structure Interaction for Building Structures.
Bray JD, Travasarou T and Zupon J (2010) Seismic displacement NEHRP Consultants Joint Venture under Project ATC-83,
design of earth retaining structures. In Proceedings of the 3rd National Institute of Science and Technology, Washington
Earth Retention Conference, ER2010, Seattle (Finno RJ, DC, USA, Report NIST GCR 12-917-21.
Hashash YMA, Arduino P and Bellevue WA (eds)). ASCE, PEER TBI (Pacific Earthquake Engineering Research Center: Tall
Reston, VA, USA, pp. 638655. Buildings Initiative) (2014) http://peer.berkeley.edu/tbi/
CTBUH (Council on Tall Buildings and Urban Habitat) (2014) (accessed 18/07/2014).
http://www.ctbuh.org/ (accessed 18/07/2014). Seed HB and Idriss IM (1970) Soil Moduli and Damping Factors
Callisto L and Soccodato FM (2010) Performance-based design of for Dynamic Response Analyses. Earthquake Engineering

40
Downloaded by [] on [12/09/16]. Copyright ICE Publishing, all rights reserved.
 Geotechnical Engineering Seismic stability of braced excavations
Volume 168 Issue GE1 next to tall buildings
ORiordan and Almufti

Research Center, University of California, Berkeley, CA, Tall Buildings Initiative. Pacific Earthquake Engineering
USA, Report No. EERC 70-10. Research Center (PEER). University of California, Berkeley,
Seed HB and Whitman RV (1970) Design of earth retaining CA, USA.
structures for dynamic loads. In Proceedings of ASCE Trombetta NW, Hutchinson TC, Mason HB et al. (2012)
Specialty Conference, Lateral Stresses in the Ground and Centrifuge modeling of structuresoilstructure interaction:
Design of Earth-Retaining Structures, Cornell, pp. 103 seismic performance of inelastic building models.
147. Proceedings of the 15th World Conference on Earthquake
Sitar N, Mikola RG and Candia G (2012) Seismically induced Engineering, Lisbon. International Association for
lateral earth pressures on retaining structures and basement Earthquake Engineering, Tokyo, Japan.
walls. In Proceedings of GeoCongress 2012: Geotechnical Ukritchon B, Whittle AJ and Sloan SW (2003) Undrained stability
Engineering State of the Art and Practice (Rollins K and of braced excavations in clay. Journal of Geotechnical and
Zekkos D (eds)). ASCE, Reston, VA, USA, Geotechnical Geoenvironmental Engineering, ASCE 129(8): 738755.
Special Publication 226. Wood JH (1973) Earthquake Induced Soil Pressures on
Stewart JP, Seed RB and Fenves GL (1998) Empirical Evaluation Structures. PhD thesis, California Institute of Technology,
of Inertial SoilStructure Interaction Effects. Pacific Pasadena, CA, USA.
Earthquake Engineering Research Center, University of Yahyai M, Mirtaheri M, Mahoutian M, Daryan AS and Assareh
California, Berkeley, CA, USA, Report PEER-98/07. MA (2008) Soilstructure interaction between two adjacent
Stewart JP and Tileylioglu S (2007) Input Ground Motions for buildings under earthquake load. American Journal of
Tall Buildings with Subterranean Levels, Task 8 Final Report Engineering and Applied Sciences 1(2): 121125.

WHAT DO YOU THINK?

To discuss this paper, please email up to 500 words to the
editor at [email protected]. Your contribution will be
forwarded to the author(s) for a reply and, if considered
appropriate by the editorial panel, will be published as a
discussion in a future issue of the journal.
Proceedings journals rely entirely on contributions sent in
by civil engineering professionals, academics and students.
Papers should be 20005000 words long (briefing papers
should be 10002000 words long), with adequate illustra-
tions and references. You can submit your paper online via
www.icevirtuallibrary.com/content/journals, where you
will also find detailed author guidelines.

41
Downloaded by [] on [12/09/16]. Copyright ICE Publishing, all rights reserved.