Bull Earthquake Eng (2011) 9:975–995

DOI 10.1007/s10518-011-9249-3

ORIGINAL RESEARCH PAPER

Seismic analysis of deep tunnels in near fault conditions:

a case study in Southern Italy

Mirko Corigliano · Laura Scandella ·

Carlo G. Lai · Roberto Paolucci

Received: 3 August 2009 / Accepted: 6 February 2011 / Published online: 5 March 2011
© Springer Science+Business Media B.V. 2011

Abstract The importance of underground structures in transportation and utility networks

makes their vulnerability to earthquakes a sensitive issue. Underground facilities are usu-
ally less vulnerable to earthquakes compared to above-ground structures, but the associated
risk may be relevant, since even a low level of damage may affect the serviceability of a
wide network. Seismic analysis of tunnels close to seismogenic faults is a complex prob-
lem, which is often neglected at the design stage for the lack of specific codes or guidelines
for the design of underground structures in seismic conditions and also because, as men-
tioned above, underground structures are considered less vulnerable to earthquake loading.
This paper investigates the seismic response of deep tunnels focusing on the required steps
for a proper design under both static and dynamic loading. The study aims at contributing
to improve the methods currently used for the seismic analysis of underground structures.
At this purpose, the seismic response of a deep tunnel in Southern Italy has been investi-
gated along the transversal direction. The infrastructure is part of the railway switch line
connecting Caserta to Foggia in the Southern Apennines which is one of the most active
seismic regions in Italy. The seismic response in the transversal direction has been analysed
by using the pseudo-static approach as well as through advanced numerical modeling using
the spectral element method coupled with a kinematic approach for finite fault simulations.
The pseudo-static approach has been implemented using a closed-form analytical solution.

M. Corigliano (B) · L. Scandella · C. G. Lai

European Centre for Training and Research in Earthquake Engineering, EUCENTRE, Via Ferrata 1,
CAP 27100, Pavia, Italy
e-mail: [email protected]
L. Scandella
e-mail: [email protected]
C. G. Lai
e-mail: [email protected]

R. Paolucci
Politecnico di Milano, Department of Structural Engineering,
P.za Leonardo da Vinci, CAP 20133, Milan, Italy
e-mail: [email protected]

123
976 Bull Earthquake Eng (2011) 9:975–995

The results obtained from advanced numerical modeling and the pseudo-static method have
been compared to assess their validity and limitations.

Keywords Deep tunnels · Seismic analysis · Soil-structure interaction ·

Near-fault conditions · Spectral element method

1 Introduction

Underground structures are critical elements in transportation and utility networks (e.g. rail-
way and road tunnels, hydraulic tunnels and hydroelectric caverns, lifelines for transportation
of water, oil, natural gas, etc.). Their seismic response is considerably different from that of
the corresponding above-ground facilities. Since the relative movement of pipes and tunnels
with respect to the surrounding soil is generally small, the structural response is dominated
by the surrounding soil behaviour. The inertia forces due to the weight of the structure
are often negligible compared to the mass of the surrounding ground, and the stress con-
finement provides high values of radiation damping. It is a fact that underground facilities
are in general less vulnerable to earthquakes in comparison with above-ground infrastruc-
tures (Barton 1984; Ates et al. 1995). Nevertheless, several tunnels and underground struc-
tures suffered severe damages, especially during strong earthquakes such as the recent 1995
Kobe (Japan), 1999 Chi-Chi (Taiwan) and 2004 Niigata (Japan) events. A review of dam-
ages suffered by underground structures during these events shows that most tunnels were
located in the vicinity of causative faults. This condition implies a high factor of risk since
ground motion may be characterized by strong and coherent long period pulses. Therefore,
a careful definition of the seismic input for the seismic assessment of these structures is
required.
The static design of tunnels nowadays has achieved a high level of refinement, mainly
due to the availability of sophisticated numerical methods which allow one to simulate all
construction phases (such as excavation), cast in place of temporary reinforcement and mod-
elling of the complex boundary conditions. A similar level of accuracy is not yet developed
for the estimation of stress increments in tunnel linings due to earthquake loading. This is
partly due to the fact that tunnels are not considered very sensitive to earthquakes, and only
few code prescriptions and guidelines specifically address the issue of seismic design (e.g.
ISO 23469 2005 and the French AFPS/AFTES 2001).
This paper aims at analyzing the seismic response of deep tunnels in near field conditions
and to introduce improved approaches for the seismic design of underground structures. This
is achieved by means of both simplified methods and advanced numerical simulations with
the goal of assessing advantages and limitations of approaches characterized by different
levels of refinement. The attention will be specifically devoted to the steps implemented in
the design of deep tunnels under earthquake loading, namely:
(i) characterization of rock mass for stress analysis under static conditions;
(ii) static analysis;
(iii) definition of seismic input in near-field conditions;
(iv) stress analysis under seismic loading;
(v) superposition of static and dynamic loading effects.
The proposed procedure will be applied to a real case study, the Serro Montefalco tunnel
located in Southern Italy. The infrastructure is part of the railway switch line connecting Ca-
serta to Foggia, in the Northern sector of the Southern Apennines, one of the most seismically

123
Bull Earthquake Eng (2011) 9:975–995 977

active regions in Italy. The paper, after a brief description of the geological and seismologi-
cal setting of the case study, illustrates the methods adopted for the definition of the seismic
input required for dynamic analyses of the tunnel considering the transverse response. Special
attention has been devoted to the evaluation of strains, which dominate the seismic response
of underground structures, such as tunnels or pipelines.
The method of Hisada and Bielak (2003) has been adopted to generate synthetic time
histories of earthquake ground motion, based on an extended kinematic fault model. The
synthetic seismograms have been used to assess tunnel response via advanced numerical
analyses as well as for the evaluation of the maximum shear deformation induced within the
tunnel, as required by simplified approaches.
Since the assessment of the seismic behaviour of a tunnel along the transversal direction
requires the knowledge of the internal forces acting on the lining prior to the earthquake, static
analyses simulating the evolution of the state of stress during all phases of construction of the
tunnel have been performed using FLAC (Itasca 2005), a well-known finite difference-based
computer program.
Subsequently, the stress increment in the lining due to the seismic excitation has been
calculated and compared using a pseudo-static approach and advanced numerical modeling.
Numerical analysis has been carried out using GeoELSE (Geo Elastodynamics by Spectral
Elements, http://geoelse.stru.polimi.it/), a spectral element-based computer program coupled
with GRFLT, a computer code developed by Hisada and Bielak (2003) for finite fault simula-
tions. The coupling has been implemented through the use of the Domain Reduction Method
(Bielak et al. 2003; Scandella 2007; Stupazzini et al. 2006). The pseudo-static approach has
been implemented using a closed-form analytical solution developed by Corigliano et al.
(2006).

2 Case study: the Serro Montefalco tunnel

The new “Caserta-Foggia” railway line is part of the doubling of the original route repre-
senting one of the most important crossings of the Apennines, in Southern Italy. The new
railway line was designed in the late 80’s and includes 17 tunnels. The line between Caserta
and Apice (close to Benevento) was built in the early 90’s whereas for the last part of it only
a preliminary design has been carried out.
The “Serro Montefalco” tunnel belongs to the final section of the “Caserta-Foggia” line.
This is a 11.7 km long tunnel with a maximum cover of 225 m. It represents one of the most
relevant structures of the entire railway line due to the complexity of the geological setting
(see Fig. 1b). The lithotypes include varicoloured clay-shales, marl and marly limestone,
clay and marl intercalated with limestone (Barla et al. 1986). The varicoloured clay-shales
(the so-called “Argille Scagliose”) include expansive clay minerals, which exhibit a signif-
icant swelling behaviour. Previous excavations of tunnels (e.g. the “San Vitale” Tunnel) in
these weak rock formations (see Barla et al. 1986 and Lunardi and Bindi 2004 for more
details) presented severe squeezing and swelling problems leading to face instability, large
convergences, invert-heave and critical loading of the tunnel support.
The “Serro Montefalco” tunnel has been chosen as case study due to the relevant seis-
micity of the region and the availability of the design data. The Southern Apennines are
characterized by a narrow seismic belt NW-SE-striking and approximately 30–50 km wide,
which follows the axis of the chain (Improta et al. 2000). The Northern sector of the Southern
Apennines, the “Sannio” region, where the “Serro Montefalco” tunnel is located, is one of
the most active seismic territories in Italy. This area in the last three centuries was struck by

123
978 Bull Earthquake Eng (2011) 9:975–995

Fig. 1 a Location of the “Serro Montefalco” tunnel (thick line) along the Caserta-Foggia railway line (gray
line) in the Sannio region. Epicentres of the most destructive earthquakes in the last three centuries (stars)
are shown. The active faults (from the DISS 3.0.2 (2006) database) close to the tunnel are superimposed; the
Ariano Irpino fault used as seismic source in the dynamic analysis is highlighted. The thin line A–A’ shows
the representative modelled cross-section of the tunnel. b Geological profile along the Caserta-Foggia line,
where the “Serro Montefalco” tunnel is located (from Barla et al. 1986)

four large destructive earthquakes occurred in 1688 (IMCS = XI), 1702 (IMCS = X), 1732
(IMCS = X), and 1805 (IMCS = X), as shown in Fig. 1a. Since 1805, a period of seismic
quiescence followed. At present, the seismicity is characterized by low-energy earthquakes
frequently clustered in swarms which occur at the borders of primary faults that are sources
of the most energetic earthquakes (Improta et al. 2000). Due to this long quiescence, the
Sannio region is expected to experience in the near future a strong and destructive seismic
event.

123
Bull Earthquake Eng (2011) 9:975–995 979

3 Seismic ground strains in near-fault conditions

It is well known that long infrastructures, such as viaducts, pipelines and tunnels, require a
careful selection of the seismic input to account for the spatial variability of ground motion,
especially in near-fault conditions, where the motion may strongly depend on the complex
coupling of the seismic source, rupture process, source-site travel path and local site condi-
tions (Stupazzini et al. 2009). Consequently, the features of ground motion in the proximity
of an active fault can be significantly different from those in the far-field. There is no well
defined distance over which a site may be classified as in near or far-field. A useful criterion
to define the near-field zone is based on the comparison of the source dimension with the
source to site distance. Experience shows that only at distances smaller than 15 ÷ 25 km
relevant damage to underground facilities is expected to occur.
Since the seismic response of underground structures is dominated by the surrounding
ground deformation and not by the inertial properties of the structure itself, the standard rep-
resentation of the seismic input in terms of response spectral ordinates is not suitable. Instead,
the dynamic analysis of the structure should be carried out using three components time histo-
ries of ground motion. According to Eurocode 8 Part 1 (EN 1998-1, 2004), acceleration time
histories can be defined using different approaches including artificial spectrum-compatible
accelerograms, synthetic records generated by a seismological model of the source or real
accelerograms. However, the use of artificial spectrum-compatible accelerograms for geo-
technical applications is not allowed by the EC8 and also by the recent Italian building code
(NTC 2008).
In spite of the availability of large digital accelerometric databases in many seismically
active regions of the world, relatively few good quality strong motion records are available in
near-fault conditions. Almost none at depth below the ground level. These would be useful
to define the seismic input for an underground structure located close to a seismogenic fault.
Therefore, to define a suitable seismic input for the problem under study, two approaches
have been adopted, namely:
– generation of synthetic time series using the method by Hisada and Bielak (2003), which
simulates the seismic source as an extended kinematic fault (see Sect. 3.1);
– simplified evaluation of the rock Peak Ground Strain (PGS) for simplified pseudo-static
analyses (see Sect. 3.2).

3.1 Seismic ground motion by finite fault numerical simulations

To provide a suitable seismic input for dynamic analyses of the tunnel, finite fault numerical
simulations have been performed using the method proposed by Hisada and Bielak (2003)
that simulates the complete 3D seismic wavefield induced by an extended kinematic fault
coupled with a horizontally layered crustal model. This approach allows to investigate the
effects of the source, fling step, rupture directivity and strong motion in near-fault conditions,
which may play a relevant role in the expected ground shaking.
Proper definitions of the seismic source model and of the crustal profile along the whole
propagation path are required for the application of the method.

3.1.1 Source model

The Italian Database of Individual Seismogenic Sources (DISS 2006) indicates three relevant
faults close to the “Serro Montefalco” tunnel: the Ariano Irpino, Ufita Valley and Tammaro

123
980 Bull Earthquake Eng (2011) 9:975–995

Table 1 Main parameters of the

Seismic moment 2.54 × 1019 Nm Strike 277◦
Ariano Irpino fault (from DISS
V3.0.3 2006) Moment magnitude 6.9 Dip 70◦
Min. depth 11 km Rake 230◦
Max. depth 25 km Length 30.0 km
Hypo. depth 22.7 km Width 14.9 km
Rise time 1.8 s Slip 2.0 m
Rupture velocity 2.8 km/s Max freq. 5 Hz

Shear wave velocity (m/s)

0 1000 2000 3000 4000
0 0 1000 2000 3000 4000
0
2000
500
4000
1000
6000
Depth (m)

8000 1500

10000 2000

12000 Cotton et al.,2006 - 2500

Site effects
14000 adjustments 3000
Crustal velocity
16000 model used for 3500
18000 Sannio region
Improta et al., 2000 4000
20000

Fig. 2 Crustal velocity profile (S wave) adopted for the numerical model

Basin faults all of which are shown in Fig. 1a. The Ariano Irpino fault has been selected
as the reference seismic source, because it is the closest to the tunnel and it is capable of a
Mw 6.9 earthquake. It was the source of the December 5, 1456 earthquake, one of the most
important natural events of the Italian seismic history. The main features of this source and
the parameters adopted for the deterministic simulation are listed in Table 1.

3.1.2 Crustal velocity model

To define a crustal velocity profile for application of the Hisada and Bielak (2003) method,
the Improta et al. (2000) model was considered for the deeper part, which is based on the
interpretation of seismic refraction data and of several sonic velocity logs obtained from oil
wells in the region surrounding the area of study. To provide a more realistic shallow velocity
profile, we have adopted within the first 5 km from ground surface the rock model proposed
by Cotton et al. (2006), assuming VS30 = 600 m/s. The resulting layered crustal model is
shown in Fig. 2 and the material properties adopted in the numerical model are listed in
Table 2.

3.2 Evaluation of ground strain field

As mentioned above, the seismic response of underground structures is dominated by

earthquake-induced deformation of the surrounding soil. Since direct measures of earth-
quake-induced ground strains are generally not available, and since the ground deformation

123
Bull Earthquake Eng (2011) 9:975–995 981

Table 2 Material properties assumed for the layered crustal model used for the Ariano Irpino fault ground
motion simulations
Layer ρ VP VS QP QS Depth Thickness
[t/m3 ] [m/s] [m/s] − − [m] [m]

1 1.80 1,039 600 60 30 from 0 to 25 25

2 1.90 1,386 800 100 50 from 25 to 50 25
3 2.10 2,252 1,300 150 75 from 50 to 150 100
4 2.30 3,031 1,750 200 100 from 150 to 250 100
5 2.35 3,464 2,000 200 100 from 250 to 400 150
6 2.40 3,724 2,150 200 100 from 400 to 500 100
7 2.40 4,070 2,350 260 130 from 500 to 1,000 500
8 2.50 4,503 2,600 300 150 from 1,000 to 1,500 500
9 2.55 4,850 2,800 320 160 from 1,500 to 2,000 500
10 2.55 5,196 3,000 320 160 from 2,000 to 3,000 1,000
11 2.60 5,543 3,200 320 160 from 3,000 to 4,000 1,000
12 2.60 5,785 3,340 320 160 from 4,000 to 5,000 1,000
13 2.70 6,062 3,500 400 200 from 5,000 to 7,000 2,000
14 2.70 6,582 3,800 500 250 from 7,000 to 11,000 4,000
15 2.55 5,023 2,900 320 160 from 11,000 to 19,150 8,450
16 2.70 6,582 3,800 500 250 from 19,450 to ∞ −

pattern in realistic geological and tectonic conditions is hard to predict, it is customary in the
engineering practice to make use of simplified formulas.
The Peak Ground Strain (PGS) is related to the Peak Ground Velocity (PGV) by means
of the following relation (Newmark 1967):
P GV
PGS = (1)
C
where PGS may refer to either the longitudinal or the shear deformation. The parameter C
(dimensionally is a velocity) is defined in terms of the apparent speed of propagation of the
dominant phases of seismic motion, either body or surface waves, with values typically rang-
ing from 2 and 5 km/s (O’Rourke 2003). As discussed in several papers (see e.g. Zerva 2003;
Paolucci and Pitilakis 2007; Paolucci and Smerzini 2008; Scandella and Paolucci 2010),
the practical application of Eq. (1) may be limited especially when the simplifying assump-
tion on which this relation is based (i.e. of a seismic wavefield consisting of plane waves
propagating in a homogeneous medium) is not applicable. Therefore, in this work we have
calculated ground strains based on the kinematic modeling of the seismic source described
in Sect. 3.1. The maximum earthquake-induced ground strain was calculated by numerical
differentiation of the computed numerical displacement time histories v(y, z) and w(y, z) at
four points around the cross-section of the tunnel (Corigliano et al. 2006), as shown in Fig. 3.
Denoting by “x” the longitudinal axis of the tunnel, the shear strain γ yz have been computed
as follows:
∂v ∂w ∼ 1
γ yz = + = [v (yo , z o + z) − v (yo , z o − z)]
∂z ∂y 2 z
1
+ [w (yo + y, z o ) − w (yo − y, z o )] (2)
2 y

123
982 Bull Earthquake Eng (2011) 9:975–995

Fig. 3 Displacement time

histories evaluated by the Hisada
and Bielak (2003) method in Displacement
free-field conditions at the tunnel in y direction
depth in 4 points around the
tunnel

Δz
Displacement
in z direction

Δy

where the partial derivatives have been computed using the second order, central finite dif-
ference operators, v and w are the ground displacement in y and z direction, y and z are
the spacings between two points considered in y and z directions.

4 Response along the transversal direction of the tunnel

The seismic response of a representative transversal cross-section of the Serro-Montefalco

tunnel has been analyzed. Special attention has been devoted to the ovaling, which is one of
the most critical deformation patterns induced by ground shaking to a tunnel lining (Hashash
et al. 2001). The estimation of the seismic stress in the lining requires a preliminary cal-
culation of the static stresses acting prior to the earthquake loading. In this study, after the
static analysis, the seismic response has been analysed by two approaches: a fully dynamic
analysis and a pseudo-static approach. In both approaches, the free-field ground strains have
been calculated as illustrated in the previous section, to provide a meaningful comparison
and a cross-validation of the two methods.
The main scope of the static analysis is the reproduction of the stress in the lining due to
the excavation and construction stages of the tunnel. On the other hand, dynamic analyses
aim at evaluating the stress increment in the lining due to the seismic wave propagation.
It is important to remark that the geotechnical parameters used under static and dynamic
loading are different, due to the different level of deformation involved. Static parameters
have been defined on the basis of the standard methods used in rock mechanics, whereas
in the dynamic analysis elastic moduli at low strains were used. In both static and dynamic
analyses, the rock mass has been modelled as an isotropic, linear elastic continuum since the
influence of discontinuities in weak rocks can be neglected.

4.1 Static analysis

A set of static analyses have been carried out to properly simulate the construction stages of
the tunnel and to compute cross-sectional internal forces in the lining prior to the earthquake

123
Bull Earthquake Eng (2011) 9:975–995 983

Table 3 Strength and

Cohesion 50 kPa
deformability parameters of
varicoloured clay-shales Friction angle 22◦
(Barla et al. 1986) Young’s modulus 200 MPa
Poisson ratio 0.45

86
6.

5.2
5

(a) (b)
Fig. 4 a Cross-section of the “Serro Montefalco” tunnel; b mesh of the FLAC model

loading. The parameters of the rock mass along the tunnel length have been defined on the
basis of geomechanical classification and scaling rules of the intact rock properties obtained
from laboratory tests.
For soil-like materials (e.g. varicoloured clay-shales) reference was made to laboratory
tests conducted on undisturbed samples (e.g. triaxial tests, direct shear tests, consolidation
tests, etc.) obtained from borehole drilling and from a limited number of field investigations
such as flat-dilatometer tests (Barla et al. 1986). The strength and deformability parameters
adopted for the varicoloured clay-shales formation are listed in Table 3.
Static analyses have been performed using the finite difference-based code FLAC (Itasca
2005). A two-dimensional model of the tunnel under plane strain conditions (Fig. 4b) has
been constructed. In order to simulate the excavation stages, which determine 3-D stress-
strain conditions near the tunnel face, the stress relaxation method proposed by Panet (1995)
has been used. This introduces a fictitious stress (σ )c acting on the two-dimensional model
as follows:

(σ )c = (1 − λ) (σ ) p (3)

where λ is the stress relaxation factor ranging between 0 and 1 depending on the distance of
the cross-section from the tunnel face, and (σ ) p is the geostatic state of stress.
Calculations have been performed with reference to a geostatic stress in the rock mass
given by an overburden pressure σv = 2.14 MPa (which corresponds to a depth of approxi-
mately 100 m) and with at-rest stress coefficient K 0 = 0.8 (Barla et al. 1986), where K 0 is
the ratio between the effective horizontal stress σh and the effective vertical stress σv .
The rock mass has been modelled using an elastic-perfectly-plastic constitutive law, based
on the Mohr-Coulomb yield criterion and a non-associated flow rule. The construction stages
have been simulated through the following steps:

123
984 Bull Earthquake Eng (2011) 9:975–995

Table 4 Mechanical parameters of the primary support

Shotcrete Thickness = 30 cm ν = 0.2 Eshot = 4,000 MPa

Steel ribs HEB200 Asteel = 78.1 cm2 D = 1.0 m Esteel = 210,000 MPa
Jsteel = 5,696 cm4
Equivalent parameters G = 3,932 MPa ν = 0.2 Eeq = 9,438 MPa
K = 5,243 MPa

Table 5 Mechanical parameters of final lining

E γ ν K G Thickness at the crown Thickness at the invert

30 GPa 25 kN/m3 0.2 16.67 GPa 12.50 GPa 80 cm 110 cm

• simulation of the initial geostatic stress;

• full face excavation up to a 50% removal of the stress on the tunnel profile;
• application of the primary support at the crown and closure of the ring by the reinforced
concrete lining installed at the invert;
• additional removal up to 75% of the initial stress on the tunnel profile, with the above
support system installed;
• installation of final concrete lining at the crown and complete relaxation of the stress on
the tunnel profile;
• degradation of the geotechnical parameters of the primary support.

The primary support consists of a ring of a 30 cm thick shotcrete layer and steel ribs (1 HEB
200 per meter). It has been modelled by plane strain elements with a linear elastic isotro-
pic behaviour, characterized by an equivalent Young’s modulus (Oreste 1999). The elastic
parameters used for the primary support are listed in Table 4.
The final reinforced concrete lining, which acts as a permanent structural support of the
tunnel, will carry both the whole static load and the additional dynamic loading increment
caused by the earthquake. A typical cross-section of the tunnel located in the varicoloured
clay-shales formation, which includes the primary and final lining, is shown in Fig. 4a. The
final reinforced concrete lining has been idealized using linear elastic, isotropic plane strain
elements (see Fig. 4b) with the elastic parameters reported in Table 5. The cross-sectional
forces acting along the lining are shown in Figs. 5 and 6.
Figure 5 shows that the state of stress in the lining is governed by the stress increment
induced during the construction stages, which includes the cast in place of the invert, the
primary support close to the tunnel face and the closure of the crown. The sequence of the
construction stages generates in the final step large bending moments at the invert with stresses
approaching the limit design values for the cross-section (see Fig. 6b). Thus, underground
structures excavated in complex geological/geotechnical settings, like the “Serro Montefal-
co” tunnel, may undergo significant damages if an additional loading increment (such as that
due to an earthquake) is superimposed (Corigliano et al. 2007).
A similar situation occurred in Turkey to the Bolu tunnel, which crosses the North
Anatolian Fault and was severely damaged due to the Mw 7.1 Düzce earthquake of November
12, 1999 (Erdik 2000).

123
Bull Earthquake Eng (2011) 9:975–995 985

(a) 3000

Internal forces [kN/m - kNm/m]

2000

1000

-1000
Thrust force
-2000 bending moment
Shear force ϑ
-3000

-4000

-5000

-6000
0 45 90 135 180 225 270 315 360
Angle ϑ [°]
(b) 10 (c) 10
Tunnel perimeter Tunnel perimeter
Bending moment Thrust force
Shear force
5 5

0 0
-10 -5 0 5 10 -10 -5 0 5 10

-5 -5

-10 -10

Fig. 5 a Cross-sectional forces acting on the reinforced concrete lining at the final stage of construction:
thrust force (triangles); bending moment (diamonds) and shear force (circles); b Bending moment and shear
force along the tunnel perimeter; c Thrust force along the tunnel perimeter

4.2 Advanced dynamic modeling

Advanced numerical analyses have been performed to assess the dynamic response of the tun-
nel cross-section during seismic loading. These include simultaneous consideration of seismic
source, propagation path, geological site conditions, the effects of dynamic soil-structure
interaction. The analyses have been carried out using GeoELSE (Geo-ELastodynamics by
Spectral Elements, http://geoelse.stru.polimi.it), a spectral element-based computer program
jointly developed at the Department of Structural Engineering of Politecnico di Milano
(Faccioli et al. 1997) and at CRS4 (Center for Advanced Studies, Research & Development
in Sardinia, Italy). GeoELSE is a powerful computer program implemented in a parallel
architecture (Stupazzini et al. 2009) that has been specifically designed to solve complex
wave propagation and dynamic soil-structure interaction problems.
GeoELSE capabilities were further enhanced by the implementation of the Domain Reduc-
tion Method (Bielak et al. 2003; Scandella 2007), a powerful substructuring approach which
reduces the computational cost required in the numerical analysis of large scale dynamic
soil-structure interaction problems. The main idea behind DRM is the subdivision of the

123
986 Bull Earthquake Eng (2011) 9:975–995

(a) Interaction diagramm Crown section 80x100

5000

Bending moment [kNm/m]

Interaction Diagram
Static Analysis
2500

0
-35000 -25000 -15000 -5000 5000

-2500

-5000
Thrust force [kN/m]

(b) Interaction diagramm Invert section 110x100

5000
Bending moment [kNm/m]

2500

0
-35000 -25000 -15000 -5000 5000

-2500

-5000
Thrust force [kN/m]
Fig. 6 Interaction diagrams (solid line with squares) for the tunnel cross-section in static condition (circles):
a crown section 80 × 100; b invert section 110 × 100

original problem into two simpler subproblems, solved in two sequential steps, as illustrated
in Fig. 7. Step 1 (see Fig. 7a) simulates the seismic response of the free-field (auxiliary)
problem, consisting of the wave propagation generated by a seismic source placed inside a
medium (external domain) from which the structure has been removed and replaced by the
same material as the surrounding soil. Step 2 (see Fig. 7b) involves the numerical solution
of the reduced problem involving the structure alone (in this case, the tunnel) and only a
reduced portion of the surrounding soil (internal domain). Therefore, the DRM allows to
perform an exact coupling between a large scale external domain, involving the seismic
source, but excluding the structure, and an internal small-scale domain, including the struc-
ture, but excluding the source of the excitation. As shown by Bielak et al. (2003), the input
for Step 2 of DRM is a set of effective nodal forces evaluated on the basis of the ground
displacement calculated in the Step 1 and applied in a strip of elements (dark boundary in
Fig. 7). These forces are equivalent to and replace the original seismic forces applied in the
first step to reproduce the seismic source. A detailed description of the implementation of
the DRM applied to Spectral Elements can be found in Faccioli et al. (2005) and Scandella
(2007).
In this study the auxiliary problem (external domain) has been solved using the GRFLT
program developed by Hisada and Bielak (2003) (see Sect. 3.1) assuming a layered crustal

123
Bull Earthquake Eng (2011) 9:975–995 987

Fig. 7 DRM procedure applied to the case study. a Step I: free-field analysis of the wave propagation from
the seismic source; b Step II: wave propagation in the reduced domain including soil-structure interaction.
The dark line denotes the effective boundary. c SE model of the reduced problem including the transversal
A–A’ cross-section of the tunnel

model with properties illustrated in Table 2. The results of the analyses are shown in Fig. 3
in terms of displacement time histories at selected nodes. As expected, the seismic motion
does not vary significantly at the various nodes around the tunnel due to the wavelengths
propagated that are large with respect to the dimensions of the structure. However, the spa-
tial variability of the displacement field induced by wave propagation effects generates a
significant deformation field, as it will be shown in the next sections.
As a second important remark, it should be noticed that, although the numerical sim-
ulation with GeoELSE was set to propagate frequencies up to 5 Hz, the modelling of the
seismic source through the GRFLT program allowed to reproduce the radiation of energy at
frequencies not exceeding 2 Hz. Therefore, the magnitude of the deformation field reproduced
by the numerical simulations could be underestimated.
Step 2 of the DRM has been solved applying GeoELSE to a 2D numerical model of the
cross-section A–A’ of the “Serro-Montefalco” tunnel shown in Fig. 1a. A circular cross-sec-
tion of the tunnel was adopted with an equivalent external radius of 5.85 m and a concrete
lining thickness of 0.80 m. This simplified the geometry for the numerical modeling and
allowed to compare the results with those obtained with the pseudo-static approach.
The Young’s modulus and Poisson’s ratio adopted for the lining are shown in Table 5. The
SEM model of the reduced problem, shown in Fig. 7c, has the mechanical properties shown
in Table 6. A no slip condition has been assumed at the interface between the soil and the
lining, as further discussed in the next section.
The maximum dynamic increment of the axial force and of the bending moment obtained
from advanced numerical modeling will be compared with the results obtained by a simplified

123
988 Bull Earthquake Eng (2011) 9:975–995

Table 6 Dynamic properties

Layer N. VS (m/s) VP (m/s) ρ (kg/m3 ) QS (–)
adopted for the reduced model in
the advances numerical analyses
1 600 1,039 1,800 30
2 800 1,386 1,900 50
3 1,300 2,252 2,100 75
Tunnel 2,236 3,651 2,500 80

closed-form solution derived using a pseudo-static approach, taking into account soil-struc-
ture interaction effects in no-slip conditions.

4.3 A closed-form solution for simplified pseudo-static analysis

A simplified pseudo-static analysis of the transversal cross-section of the tunnel has been
performed applying the closed form solution developed by Corigliano et al. (2006). The
approach considers a lined circular tunnel (diameter 5.85 m and thickness 0.80 m) under
plane strain conditions. The rock mass is considered to be an infinite, linear-elastic, homoge-
neous, isotropic medium. The tunnel lining is modeled as an elastic cylindrical shell in plane
strain conditions.
Since the real boundary conditions at the soil-structure interface are unknown, simplified
solutions available in the literature for the evaluation of seismic actions generally consider
two extreme cases which bound the real situation: the full-slip and the no-slip conditions. In
the engineering practice the two analytical solutions proposed by Wang (1993) and Penzien
(2000) are widely used for the evaluation of the dynamic stress increment in the tunnel lin-
ing. The two solutions provide the same result for the full-slip condition, but the solution
proposed by Penzien (2000) underestimates the thrust force in the no-slip case. This aspect
was observed by Hashash et al. (2005) while studying the problem with Finite Elements
(FE) numerical analyses and it was confirmed also by Kontoe et al. (2008), who compared
FE numerical results with the simplified analytical solutions calculated for the Bolu tunnels,
which experienced extensive damage after the 1999 Düzce earthquake. On the other hand,
Hashash et al. (2005) showed that the solution by Wang (1993) leads to correct results. This
has also been confirmed by the analytical solutions independently developed by Corigliano
et al. (2006) which will be briefly recalled hereinafter.
A full-slip condition is usually adopted to obtain the extreme values of the bending moment
and shear force in the tunnel lining, whereas the no-slip assumption yields the maximum val-
ues of the thrust force acting on the lining. Since the shear force and the bending moment in
full and no-slip conditions are slightly different (10 ÷ 20%), while the thrust in no-slip con-
dition can be much higher than in full-slip (100 ÷ 200 times), for the sake of conservationism
the no-slip condition has just been assumed as the reference boundary condition.
The earthquake loading has been modeled as a uniform, quasi-static strain field simulating
a pure shear deformation. This choice is justified by two reasons: firstly, the size of a typical
lining cross-section is small compared with the wavelengths of the dominant ground motion
producing the ovaling; secondly, the inertia effects in both the lining and the surrounding
ground as produced by dynamic soil-structure interaction effects are relatively small (Penzien
2000). In an infinite medium, a strain field associated to a pure shear deformation can be
obtained by applying a state of stress corresponding to an at-rest coefficient of lateral pressure
K 0 equal to −1, as shown in Fig. 8.
The relations for displacements, bending moments, thrust and shear forces have been
derived following the same approach used by Einstein and Schwartz (1979). However, the

123
Bull Earthquake Eng (2011) 9:975–995 989

τ σ

45°
ϑ
45°

Ko = -1
Ovaling of the
Lining cross section

Fig. 8 State of stress around a tunnel corresponding to uniform, pure shear deformation

Fig. 9 Internal forces acting on Mϑ

the cylindrical shell in plane
strain conditions
Qϑ Cylindrical
Nϑ

assumption that the induced internal forces are caused by excavation has been removed and
replaced with an imposed, external quasi-static loading distribution to simulate the earth-
quake loading. The relationships derived for the thrust force (Nϑ ) and bending moment
(Mϑ ) per unit length of tunnel lining (see Fig. 9) associated with the no-slip condition are
the followings (Corigliano et al. 2006, Corigliano 2007):
   
Eg δ π 
Nϑ =   γ f f max R 1 − cos 2 ϑ + (4)
2 1 + νg 3 4
   
Eg R2 δ π 
Mϑ =   γ f f max 1 + + ε cos 2 ϑ + . (5)
2 1 + νg 2 3 4
In case of full slip condition these relations become respectively:
Eg   π 
Nϑ =   γ f f max R (1 − 2η) cos 2 ϑ + (6)
2 1 + νg 4
Eg   π 
Mϑ =   γ f f max R 2 (1 − 2η) cos 2 ϑ + (7)
2 1 + νg 4
where the parameters ε, δ, η are defined by:

   ∗

2a 1 + C ∗ 1 − νg − 6 CF ∗ [a + 4]
ε=   ∗       (8)
a 4 νg − 4 − a + 3 CF ∗ 4 νg + a − 6 1 − νg [2 + a] + 1 − 2νg a
 
with: a = C ∗ 1 − νg ;
 
a − 2 − 4 νg + a ε
δ= (9)
2+a
 ∗   
F 1 − νg + 6 1/2 − νg
η =  ∗    (10)
2F 1 − νg + 6 5 − 6νg

123
990 Bull Earthquake Eng (2011) 9:975–995

(a) (b)
Bending moment [kNm/m
3000 10000
9000

Thrust force [kN/m]

2500 8000
2000 7000
6000
1500 5000
4000
1000 3000
2000 Full slip No slip
500 Full slip No slip
1000
0 0
0 100 200 300 400 0 100 200 300 400
*
F F*
Fig. 10 Effect of the contact condition at the rock-mass-lining interfaces (full slip (solid line) and no slip
(dashed line)) on the increment of the internal forces on the lining under seismic loading versus the flexibility
ratio F*: a bending moment; b thrust force

where C ∗ and F ∗ are:

   
E g R 1 − νs2
∗ E g R 3 1 − νs2
∗
C =   F =  . (11)
E s As 1 − νg2 E s Is 1 − νg2

R is the average tunnel radius, As and Is are the area and the moment of inertia per unit
length of the lining, respectively, E g , E s , νg and νs are the Young’s modulus and Poisson’s
ratio of ground and lining respectively. The parameters C ∗ and F ∗ are the “compressibil-
ity” and “flexibility” ratios. They represent a measure of the relative stiffness of the ground
with respect to the supporting system (i.e. the lining) under a symmetric and antisymmetric
loading respectively (Einstein and Schwartz 1979). Finally, f f max is the absolute value of
maximum shear strain calculated in free-field conditions.
Closed-form solutions have the advantage to reflect the parameters that control the
response of the system. Equations (4) through (7) show that the seismic response of the
cross-section depends mainly on the relative stiffness of the lining with respect to the ground
(through the parameter F ∗ for the earthquake loading), on the boundary condition at the
ground support interfaces and on the maximum free-field shear deformation.
Figure 10 shows the effect of the relative stiffness and boundary conditions in terms
of bending moment and thrust force, as function of the parameters F ∗ . From Fig. 10 it
appears that the full-slip represents a slightly more critical boundary condition for the bend-
ing moment, whereas the no-slip provides the most critical assumption for the thrust force.
It is also apparent from Fig. 10 that the full-slip condition under simple shear may cause
significant underestimation of the maximum thrust force. Particularly, slip at the interface
may be possible for tunnels in soft soils or in cases of severe seismic shaking. Since the real
contact will be at an intermediate situation between the full-slip and the no-slip conditions,
both cases should be investigated (Hashash et al. 2001).
Once the maximum shear strain induced during the seismic shaking (γ f f max ) is known,
the shear stress imposed at the boundary of the model can be easily computed using the
theory of elasticity:
Eg
σ =τ =   γ f f max (12)
2 1 + νg
where E g and νg are the elastic parameters of the rock mass.

123
Bull Earthquake Eng (2011) 9:975–995 991

The key parameter for the definition of the state of stress in the tunnel lining is the free-field
maximum shear strain. For the present case study, the earthquake-induced shear strain his-
tory γ yz was calculated using Eq.( 2), where displacement time histories u(x, y) and v(x, y)
have been computed at four points along the lining of the cross-section (see Fig. 3) using the
dynamic approach illustrated in Sect. 3.1. The maximum value of shear strain for free-field
conditions turned out to be γ f f max = 1.39 × 10−4 .
As a remark, it is noted that a simpler calculation of the maximum shear strain could have
been performed using the simplified solution for earthquake-induced ground strain proposed
by Newmark (1967) which relates the peak ground strain (PGS) to the peak particle velocity
(PGV) through the formula P G S = P GV /C (see Eq. 2). PGV has been computed using a
ground motion prediction equation specifically developed for near-field conditions (e.g. Bray
and Rodriguez-Marek 2004). The values obtained for PGV range from about 50–55 cm/s.
Typical values for C range between 2 and 4 km/s (Abrahamson 2003). Setting this parameter
to the lowest limit of 2 km/s, a value of P G S = 1.30 × 10−4 was obtained which is in very
good agreement with the value predicted by the numerical simulations carried out suing the
GRFLT program (P G S = 1.39 × 10−4 ).
Such a good agreement of the two solutions may be explained by the fact that the defor-
mation field in this case-study generated by the earthquake scenario is dominated by wave
propagation effects. A more complex rupture process would likely induce a larger degree of
incoherence of ground motion, and consequently larger values of transient ground deforma-
tion.
A reasonable agreement has also been obtained between the closed-form solutions of
the internal forces illustrated above and the results of numerical simulations (see Sect. 4.2).
Figures 11a and b show the comparison in terms of bending moment and thrust force respec-
tively. The thrust force calculated through the numerical analysis exceeds of about 30% the
one computed analytically. The satisfactory agreement between the two solutions confirms
that in near-fault conditions the inertial action is not relevant since the wavelength associated
with the ground motion is larger than the cross-section of the tunnel.
It should be remarked that the bending moment and the thrust force shown in Fig. 11
represent the dynamic contribution. The total bending moment and thrust force are obtained
by adding the dynamic increments to the static values as shown in Fig. 12. Theoretically, this
linear superposition between static and dynamic phases is incorrect since the static phase may
lead to concentration of stresses and non linear effects in some parts of the lining. However,
it has been considered a reasonable assumption to compare the dynamic contribution with
respect to the static case.
It turns out that, as a result of the severe earthquake scenario that it was assumed (it cor-
responds to a minimum recurrence period of about 2,000 years), the increase of the internal
forces on the lining due to the seismic excitation almost leads to failure of the structural
support (see Fig. 12). This outcome needs to be evidently taken into account when designing
the lining of the tunnel. Thus, a proper design of a reinforced lined tunnel in seismically
active regions requires a careful computation of the dynamic increment of the internal forces
on the lining.

5 Concluding remarks

This paper attempted to provide a contribution into a better understanding of the seismic
response of deep tunnels located in the vicinity of seismogenic faults using different methods
of analysis. A specific case-study has been used to test and compare the various approaches.

123
992 Bull Earthquake Eng (2011) 9:975–995

(a) (b)
Bending moment [kNm/m]
200 3000
150

Thrust force [kN/m]

2000
100
1000
50
0 0
-50
-1000
-100
-2000
-150 Analytical solution
Numerical solution
-200 -3000
0 45 90 135 180 225 270 315 360 0 45 90 135 180 225 270 315 360
Angle [°] Angle [°]

(c) 10 (d) 10
Tunnel perimeter Tunnel perimeter
Thrust force Bending moment
Numerical solution 5 5
Numerical solution

0 0
-15 -10 -5 0 5 10 15 -10 -5 0 5 10

-5 -5

-10 -10

Fig. 11 Comparison between dynamic increment of internal forces evaluated by the numerical (rigorous
(squares)) and analytical (simplified (solid line)) methods for the examined cross-section of the Serro Montef-
alco tunnel: a thrust force; b bending moment; c thrust force along the perimeter; d bending moment along
the perimeter

This refers to a railway tunnel located in Southern Italy in a very active seismic territory. The
static analysis of this underground structure predicts large values of stresses in the lining.
As a result the tunnel is potentially vulnerable to the earthquake loading and the accuracy of
the approach to determine the seismic stress increment is of paramount importance. Thus, a
comprehensive study involving simulation of the effects of seismic source and propagation
path accounting for near-source geological conditions and soil-structure interaction has been
performed along the transversal direction of the underground structure by using the GeoELSE
code, coupled with the Domain Reduction Method. Although the complexity of the problem
has been increased, involving a large number of assumptions, from the characteristics of the
seismic source, to the static and dynamic properties of the soil and the lining, this dynamic
approach allowed us to check the reliability of simplified pseudo-static solutions.
Due to the large stiffness of the geomaterials where the tunnel is located, the propagating
wavefield is characterized by wavelengths that are large with respect to the size of the lining.
Under these conditions soil-structure interaction effects are negligible. The same conclusion
does not generally apply to other tunnel configurations, such as shallow cut and cover tunnels,
where the softer materials and the size of the cross-section may contribute to significant SSI
effects.
A good agreement has been obtained from the comparison of the dynamic internal forces
with those calculated using a pseudo-static simplified approach.

123
Bull Earthquake Eng (2011) 9:975–995 993

(a) Interaction diagram Crown section 80x100

5000
Interaction diagram

Bending moment [kNm/m]

Static Analysis
Dynamic Analysis + 2500
Dynamic Analysis -

0
-35000 -25000 -15000 -5000 5000

-2500

-5000
Thrust Force [kN/m]

(b) Interaction diagram Invert section 110x100

5000
Bending moment [kNm/m]

2500

0
-35000 -25000 -15000 -5000 5000

-2500

-5000
Thrust force [kN/m]
Fig. 12 Interaction diagram (solid line with squares) of the transversal cross-section of the Serro Montefalco
tunnel at crown (a) and at invert (b), adding the dynamic stress increment as both a positive (squares) and
negative (triangles) contributions to the static contribution (circles)

From the analyses along the transversal directions, it turned out that the estimation of the
dynamic increment of the internal forces in the lining can be obtained in a simplified way by
estimating the earthquake-induced ground strain assuming free-field conditions.
From this case study it can be concluded that simplified approaches for the seismic
analysis of deep tunnels give reasonable results from an engineering point of view. Stud-
ies are currently under way to improve the reliability of simplified methods to compute
the earthquake-induced maximum ground shear strain under free-field conditions to fur-
ther facilitate the use of the pseudo-static approach in the seismic analyses of underground
structures.

Acknowledgments The work presented in this paper was carried out with the financial support of the Depart-
ment of Civil Protection (DPC) of Italian Government within the framework “Reluis-DPC—Linea di Ricerca
n. 6.2—Geotecnica Sismica—Costruzioni in Sotterraneo, Gallerie e Caverne in Roccia” and by the Italian
Ministry for Research and Higher Education (MiUR—Ministero dell’Università e della Ricerca) through the
FIRB Project No. RBIN047WCL (Assessment and Reduction of Seismic Risk to Large Infrastructural Sys-
tems) whose contribution is greatly acknowledged. The authors wish to thank the coordinator of the line 6.2
of the Reluis project Prof. Giovanni Barla (Politecnico di Torino)

123
994 Bull Earthquake Eng (2011) 9:975–995

References

Abrahamson NA (2003) Model for strains for transient ground motion. In: Proceedings of the workshop on the
effects of earthquake-induced transient ground surface deformations at-grade improvements. CUREE.
N. EDA-04, Oakland, CA, USA. http://www.curee.org/projects/EDA/docs/CUREE-EDA04.pdf
AFPS/AFTES (2001) Earthquake design and protection of underground structures
Ates Y, Bruneau D, Ridgway WR (1995) An evaluation of potential effects of seismic events on a used fuel
disposal fault. AECL TR-623 86 p, AECL
Barla G, Caruso G, Rondini G (1986) La problematica delle gallerie nel raddoppio della linea ferroviaria
Caserta-Foggia. Congresso internazionale su grandi opere sotterranee, Firenze, Italy
Barton N (1984) Effects of rock mass deformation on tunnel performance is seismic regions. Adv Tunn Tech
Subsurf Use 4:89–99
Bielak J, Loukakis K, Hisada Y, Yoshimura C (2003) Domain reduction method for three-dimensional earth-
quake modeling in localized regions part I: theory. BSSA 93(2):817–824
Bray JD, Rodriguez-Marek A (2004) Characterization of forward-directivity ground motions in the near fault-
region. Soil Dyn Earthq Eng 24:815–828
Corigliano M, Scandella L, Barla G, Lai CG, Paolucci R (2007) Seismic analysis of deep tunnels in rock:
a case study in Southern Ital. In: 4th international conference on earthquake geotechnical engineering.
Paper 1616
Corigliano M (2007) Seismic response of deep tunnels in near-fault conditions. PhD dissertation, Politecnico
di Torino, Italy, p 222
Corigliano M, Lai CG, Barla G (2006) Seismic response of rock tunnels in near-fault conditions. In: 1st
European conference on earthquake engineering and seismology. September 3–8, Geneva, Switzerland,
Paper 998
Cotton F, Scherbaum F, Bommer JJ, Bungum H (2006) Criteria for selecting and adjusting ground-motion
models for specific target regions: application to Central Europe and rock sites. J Seismol 10:137–156
DISS v. 3.0.3 (2006) Database of individual seismogenic sources: a compilation of potential sources for
earthquakes larger than M 5.5 in Italy and surrounding areas. http://www.ingv.it/DISS/
Einstein HH, Schwartz CW (1979) Simplified analysis for tunnel support. J Geotech Eng Div 105:499–518
Erdik M (2000) Report on 1999 Kocaeli and DŸzce (Turkey) earthquakes. In: Proceedings of the second Euro-
conference on global change and catastrophe risk management: earthquake risks in Europe, Laxenburg,
Austria. Reference: http://www.iiasa.ac.at/Research/RMS/july2000/Papers/erdik.pdf
Eurocode 8 (2004) Design of structures for earthquake resistance. Part 1: general rules, seismic actions and
rules for building. EN 1998-1, December 2004, Comité Européen de normalisation, Brussels
Faccioli E, Maggio F, Paolucci R, Quarteroni A (1997) 2D and 3D elastic wave propagation by a pseudo-
spectral domain decomposition method. J Seismol 1:237–251
Faccioli E, Vanini M, Paolucci R, Stupazzini M (2005) Comment on “Domain reduction method for three-
dimensional earthquake modeling in localized regions, part I: theroy,” by J. Bielak, K. Loukakis,
Y. Hisada, and C. Yoshimura, and “part II: verification and applications,” by C. Yoshimura, J. Bielak, Y.
Hisada, and A. Fernández. Bull Seism Soc Am 95(2):763–769
Hashash YMA, Hook JJ, Schmidt B, Yao JIC (2001) Seismic design and analysis of underground structures.
Tunn Undergr Space Technol 16:247–293
Hashash YMA, Park D, Yao JIC (2005) Ovaling deformations of circular tunnels under seismic loading, an
update on seismic design and analysis of underground structures. Tunn Undergr Space Technol 20:
435–441
Hisada Y, Bielak J (2003) A theoretical method for computing near fault ground motion in a layered half-
spaces considering static offset due to surface faulting, with a physical interpretation of fling step and
rupture directivity. Bull Seism Soc Am 93(3):1154–1168
Improta L, Iannaccone G, Capuano P, Zollo A, Scandone P (2000) Inferences on the upper crustal structure of
Southern Apennines (Italy) from seismic refraction investigations and subsurface data. Tectonophysics
317:273–297
ISO 23469 (2005) Bases for design of structures—seismic actions for designing geotechnical works
Itasca (2005) FLAC (Fast Lagrangian analysis of continua): user’s guide. Itasca Consulting Group, Inc., Min-
neapolis
Kontoe S, Zdravkovic L, Potts DM, Menkiti CO (2008) Case study on seismic tunnel response. Can Geotech
J 45:1743–1764
Lunardi P, Bindi R (2004) The evolution of reinforcement of the advanced core using fibre-glass elements.
FELSBAU J Eng Geol Geomech Tunn 4:8–19

123
Bull Earthquake Eng (2011) 9:975–995 995

Newmark NM (1967) Problems in wave propagation in soil and rocks. In: Proceedings of the international
symposium on wave propagation and dynamic properties of earth materials. University of New Mexico
Press, pp 7–26
NTC—Norme Tecniche per le Costruzioni—DM 14/01/2008
O’Rourke MJ (2003) Buried pipelines, earthquake engineering handbook charter 23. CRC Press, Boca Raton
Oreste PP (1999) Aspetti notevoli dell’analisi e dimensionamento dei sostegni di gallerie. Gallerie e opere in
sotterraneo
Panet M (1995) Le Calcul des Tunnels par la Méthode Convergence-Confinement. ENPC, Paris
Paolucci R, Pitilakis K (2007) Seismic risk assessment of underground structures under transient ground
deformations. In: Earthquake geotechnical engineering, 4th international conference on earthquake geo-
technical engineering—invited lectures, Thessaloniki, Greece. Springer, Berlin, pp 433–459
Paolucci R, Smerzini C (2008) Earthquake-induced transient round strains from dense seismic networks.
Earthq Spectra 24(2):453–470
Penzien J (2000) Seismically induced racking of tunnel linings. Earthq Eng Struct Dyn 29:683–691
Scandella L, Paolucci R (2010) Earthquake induced ground strains in the presence of strong lateral soil
heterogeneities. Bull Earthq Eng. Published on line, 23 May 2010
Scandella L (2007) Numerical evaluation of transient ground strains for the seismic response analysis of
underground structures. PhD Thesis, Politecnico di Milano
Stupazzini M, Paolucci R, Igel H (2009) Near-fault earthquake ground motion simulation in the Grenoble
Valley by a high-performance spectral element code. Bull Seismolog Soc Am 99(1):286–301
Stupazzini M, Paolucci R, Scandella L, Vanini M (2006) From the seismic source to the structural response:
advanced modelling by the spectral element method. In: 1st European conference of earthquake engi-
neering and seismology, Genève, September 2006
Wang JN (1993) Seismic design of tunnels: a state-of-the-art approach. Parsons Brinckerhoff Quade e Douglas,
Inc., New York, Monograph 7
Zerva A (2003) Transient ground strains: estimation, modeling and simulation. In: Workshop Proceedings on
the effect of earthquake-induced transient ground surface deformations on at-grade improvements, May
28 2003, Oakland, CA, USA

123