Indian Geotech J

https://doi.org/10.1007/s40098-021-00523-w

ORIGINAL PAPER

Some Aspects of Seismic Soil–Structure Interaction of Lifeline

Structures
M. N. Viladkar1 • Manendra Singh2

Received: 14 August 2020 / Accepted: 17 February 2021

Ó Indian Geotechnical Society 2021

Abstract Historically, underground structures were con- Chamoli earthquake of Uttarakhand, actual three-dimen-
sidered to be less vulnerable to earthquakes. However, sional analyses, and some liquefaction studies.
some of the recent earthquakes have demonstrated that
underground structures too can suffer severe damages, Keywords Response spectra compatible time history 
especially when these are located in the vicinity of cau- Peak ground acceleration  Three-dimensional analysis 
sative faults. In case of a shallow underground tunnel Liquefaction  Plaxis
excavated in an urban area, design and construction is
demanding due to interaction between the tunnels and the
overlying pre-existing structures. Selection of a realistic Introduction
ground motion of an earthquake plays a crucial role in the
seismic analysis and design of underground structures. In Lifeline structures, like tunnels used in metropolitan cities
the absence of realistic earthquake data for the given area, for mass rapid transit system, water conductor systems of
it becomes essential to generate artificial or synthetic time hydro power projects, roadway/railway tunnels in hilly
history compatible with the largest earthquake expected in areas, structures like national highways particularly in hill
that area. Depending upon the situation, it may become areas, gas transportation pipelines, etc., are the strategic
necessary to consider three-dimensional aspects of the elements in transportation and utility networks. Histori-
problem rather than depending on only a two-dimensional cally, such underground structures were considered to be
behavior. Strong ground shaking can as well cause loss of less vulnerable to earthquakes as compared to structures
strength in saturated cohesion less soils resulting in lique- built on the ground surface. However, underground struc-
faction. Liquefaction can cause the ground surrounding the tures cannot be treated as completely exempt to the effects
tunnels to shift, with potentially severe consequences. An of ground shaking, as was demonstrated by the 1995 Kobe
attempt has been made in this paper to discuss various earthquake in Japan, 1999 Chi-Chi earthquake in Taiwan,
analysis and design considerations of underground lifeline and 2004 Niigata earthquake again in Japan, where several
structures and then look in to the aspects of stability of underground structures suffered severe damages [1]. The
metro underground tunnels of Delhi city on basis of growing need in recent years to enlarge these transportation
response spectra compatible time histories of 1999 networks is the cause of a renewed interest for studying the
vulnerability of such underground facilities to seismic
loading. The importance of these structures makes their
vulnerability to earthquakes a very sensitive issue. A large
& M. N. Viladkar earthquake would not only cause the direct but also the
[email protected]
indirect damages resulting in not only the potential loss of
1
Department of Civil Engineering, IIT Roorkee, Roorkee, human lives but also into damage to many other infras-
India tructures. This can finally result into severe economic
2
Department of Civil Engineering, NIT Hamirpur, Hamirpur, losses, especially in view of the time that is required to
H.P., India restore the functionality of the network. In order to reduce

123
 Indian Geotech J

the potential loss of serviceability of these structures, it is sensitive to earthquakes. As a consequence, engineers often
essential to reduce the possible associated damage and the omit the evaluation of structural performance of tunnels
effects of damage. under seismic conditions at the design stage.
A careful review of the seismic damages suffered by
underground facilities shows that most tunnels were loca-
ted in the vicinity of causative faults. The characteristics of Design Considerations
ground motion in the vicinity of the source can be signif-
icantly different from that of the far-field. The ground In case of an underground project which is located at a
motion close to an active fault may be characterized by relatively shallow depth in an urban area, design and
strong, coherent (narrow band) long period pulses and is construction will probably be more demanding due to
severely affected by the rupture mechanism, the direction interaction between the underground project and the
of rupture propagation relative to the site, and the possible overlying pre-existing structure(s) such as buildings and
permanent ground displacements resulting from the fault bridges. This interaction may be more complicated, since
slip. One of the first compilations of data on damage to 71 the ‘‘effect of local site conditions’’ can play a major role.
tunnels in rock due to earthquake shaking [2] and com- This was realized during the 1995 Kobe earthquake in
parison of their behavior with estimated PGAs and PGVs Japan (damage to bridges) and 1999 Chi-Chi earthquake in
concluded that (1) collapse of tunnels due to ground Taiwan (extensive failures in tunnels and buried gas
shaking occurs only under extreme conditions, (2) damage pipelines) [1].
did not occur so long as the PGA was lower than 0.19 g Seismic response of tunnels and underground structures
and/or PGV was lower than 0.2 m/s, (3) minor to moderate is considerably different from that of above-ground facili-
damage occurred when PGA went up to 0.5 g and PGVs up ties because the overall mass of the structure is usually
to 0.9 m/s, (4) moderate to heavy damage occurred when small as compared to the mass of the surrounding soil/rock
the PGA was higher than 0.5 g, and (5) tunnel collapse and the overall confinement provides high level of radia-
occurred only when it was associated with movement of an tion damping. Therefore, seismic response of an under-
intersecting fault. ground structure is basically controlled by the response of
Another important aspect of the problem is that strong the surrounding ground and by the imposed ground
ground shaking can cause loss of strength in saturated deformation and not by the inertial characteristics of the
cohesion less soils which gives rise to phenomenon of structure itself, because the response to such an event is
liquefaction. The consequences of liquefaction may include substantially dependent on the induced ground
bearing capacity failure giving rise to sinking of a struc- deformation.
ture, lateral spreading, and the slope instability. Slope Selection of realistic ground motion of an earthquake is
instability at portals and fault displacements can cause at therefore very important for the seismic analysis and
times catastrophic damage. Liquefaction can cause the design of underground structures. The time history of an
ground surrounding the tunnels to shift, with potentially earthquake should therefore be such that it can match the
severe consequences. It can lead to collapse of multiple expected earthquake in that particular zone or at the site in
underground structures (collapse of a subway station in question, and hence, it necessitates generation of response
Kobe, Japan, during the 1995 deadly Great Hanshin spectra compatible time history.
earthquake). Ground movements associated with landslides In majority of the cases, it may not be possible to have
can cause damage to a highway facility, generate other strong motion records at a given site. Even if such records
types of secondary impacts, and rupture the water or gas are available, there is no basis to expect that a future
supply pipelines. Fire or explosions have historically been earthquake might generate the same or similar ground
a major source of damage following the earthquakes. motion. The selected time history should adequately rep-
Rupture of a gas pipe line and electric power supply lines is resent the ground motion that can be expected at the site,
often the cause of dramatic explosions and fires. Major fire and in particular, the motion that would make the structure
that broke out as a result of the 1994 Northridge earthquake respond to the highest damage potential. It is therefore
in the USA was due to the rupture of a gas pipe. essential that for predicting the earthquake response of a
The seismic analysis of underground structures is structure, synthetic time histories are generated for the
therefore a complex task, since it involves the interaction specific sites. However, there are several uncertainties in
with several disciplines including structural geology, this process, and to overcome these, decisions are required
seismo-tectonics, engineering seismology, soil and rock to be taken in a scientific and purposeful manner. Some of
dynamics, and structural dynamics. This area has by far the issues involved in the process of generation of site
remained somewhat neglected primarily because of the specific synthetic accelerograms are (1) shape of design
conventional notion that underground structures are not response spectra, (2) zero period acceleration (ZPA), (3)

123
Indian Geotech J

duration of record, (4) strong motion and decay time (en- whereas for DBE, these values are found to be half of those
velop function of time history), (5) phase characteristic of for MCE (Fig. 2).
the record, (6) number of zero crossings, and vii) realistic
derived velocity and displacement history, etc. Selection of Adequate Time History
To generate site specific time histories, first a target
response spectra and duration of time history for the site Since no major earthquake has occurred in Delhi city,
are estimated. These are generally based on specific site therefore 1999 Chamoli earthquake of lower Himalaya has
characteristics and size as well as the distance of site from been chosen for analysis. (PGA for this earthquake is larger
the epicenter of the postulated earthquake. Subsequently, a than that for 1991 Uttarkashi earthquake.) The magnitude
spectrum compatible time history is generated. Theoreti- of the earthquake was 6.8 on Richter scale. Time history
cally, there can be infinite number of compatible time record of the horizontal component of this earthquake, Ax
histories for a given response spectra and therefore, ques- after applying the base line correction, is presented in
tion often arises as to which time history has the maximum Fig. 3. Peak ground acceleration (PGA) of this earthquake
damage potential and has the near-fault effects. There is was 3.53 m/s2 which is equivalent to 0.359 of g.
therefore a need to have a spectrum compatible time his-
tory which has the near-fault effects. Several time histories Response Spectra Compatible Time History
recorded very close to the epicenter show distinguishable
velocity pulse (commonly called as ‘‘fling’’) of significant 1999 Chamoli earthquake occurred in Chamoli district of
duration, which implies that very large energy is concen- the Garhwal Himalaya which exists in zone-V of earth-
trated in that duration of the pulse. It is therefore a matter quake zoning map of India [21]. However, Delhi falls in
of research whether such time histories are more damaging zone-IV in this map. Therefore, artificial time history has to
than the other time histories, which do not have such a be generated for zone-IV. Taking the magnitude of accel-
velocity pulse. But surely, a need arises to generate spec- eration from target spectra (Figs. 1, 2) and phase from
trum compatible time histories, which have a velocity given input history (Fig. 3), response spectra compatible
pulse. It has been found that while generating spectrum time histories have been generated using Spec3, Wavgen,
compatible time history, if one uses the phase of a recorded and SeismoMatch softwares [3]. Figures 4, 5, and 6 show
time history, which has a distinguishable velocity pulse respectively, the modified time histories considering target
(instead of a random phase as is the general practice), then response spectra of MCE by Spec3, Wavgen, and Seis-
the resulting spectrum compatible time history may also moMatch softwares, respectively. Similarly, modified time
have a similar velocity pulse [3]. histories were also generated considering the target
response spectra of DBE by Spec3, Wavgen, and Seis-
moMatch softwares, respectively. The phase and time
Generation of Response Spectra Compatible Time interval of modified time history are same as the actual
History for Delhi City earthquake [3, 14, 20].
The maximum values of PGA obtained from different
Various authors, namely [4–20], have also studied the software packages are summarized in Table 1 for the sake
influence of response spectra compatible time history. of comparison. It can be seen that PGA values of the
artificial time histories obtained by Spec3, Wavgen, and
Generation of Target Response Spectra SeismoMatch increase with reduction in damping. It can be
noticed from Fig. 7 that the trend of artificial time history
As a first step, target response spectra of Delhi city have generated by SeismoMatch is quite similar to the actual
been obtained. Since Delhi city is in seismic zone-IV, earthquake (Fig. 3) except for the value of peak ground
target response spectra for this zone are to be generated. acceleration.
Target response spectra have been generated according to
[21] for Design Basis Earthquake (DBE) and Maximum
Considered Earthquake (MCE). These spectra were gen- Case History of Delhi Metro Underground Tunnels
erated for different damping ratios of Delhi silt. Plot of
spectral acceleration, Sa (in terms of g) with the time In this section, a typical section of DMRC (Delhi Metro
period T (sec), is presented in Figs. 1 and 2 for MCE and Rail Corporation) tunnels constructed in Connaught Place
DBE, respectively. Maxima values of spectral acceleration between Rajiv Square and Patel Square has been consid-
considering MCE are 0.6 g, 0.48 g, and 0.42 corresponding ered for analysis. This section of tunnels, referenced as
to 5%, 10%, and 15% damping, respectively (Fig. 1), line-B6, is situated on the Yellow line and was constructed
in Phase-I of the work of DMRC. The diameter of DMRC

123
 Indian Geotech J

Fig. 1 Response spectra for

zone IV (medium soil) for
different damping ratios, and for
maximum considered
earthquake (MCE) [20]

Fig. 2 Response spectra for

zone IV (medium soil) for
different damping ratios, and for
design basis earthquake (DBE)
[20]

Fig. 3 Horizontal acceleration

(Ax) time history of 1999
Chamoli earthquake [20]

tunnel in Connaught place is 6.26 m with an overburden study, engineering properties of alluvium have been
depth of 16.87 m. Reinforced concrete (RC) liners, with a adopted from [22, 23] and the variation of elastic modulus
thickness of 0.28 m, have been used as a permanent sup- of Delhi silt with depth is summarized in Table 2. In situ
port system. Elastic modulus of RC liners, Ec, is unit weight, cbulk and saturated unit weight, csat of Delhi silt
3.16 9 107 kPa, and the Poisson’s ratio is 0.15. Figure 8 were found to be 18 kN/m3 and 20 kN/m3, respectively. No
shows the geometry of the soil–tunnel system along with water table was encountered during tunnel excavation.
various material properties. Thickness of each soil layer was 10 m except layer 3 which
DMRC tunnels have been excavated through alluvium has a thickness of 15 m.
deposits, generally known as Delhi silt. For the present

123
Indian Geotech J

Fig. 4 Response spectra

compatible time history for
MCE by Spec3 [3, 20]

Fig. 5 Response spectra

compatible time history for
MCE by Wavgen [3, 20]

Fig. 6 Response spectra

compatible time history for
MCE by SeismoMatch [3, 20]

Two-dimensional plane-strain finite element analysis No-slip condition (perfect bond between RC liners and the
has been carried out with rectangular domain, surrounding soil mass) was assumed between the tunnel
200 m 9 60 m, using 6-noded triangular elements for the and surrounding soil. For static response, nodes along
soil mass. Elasto-plastic behavior of soil mass was simu- vertical boundaries were restrained in X-direction and were
lated by Mohr–Coulomb yield criterion. The RC liners of free to move in Y-direction, whereas the bottom boundary
the tunnel were modeled through 32 plate bending ele- was restrained in both X- and Y-directions. For dynamic
ments. Linear elastic behavior of RC liners was considered analysis, viscous absorbent boundary, proposed earlier by
in the analysis. Damping in RC liner has been taken as 2%. [24], was used to represent displacement condition along

123
 Indian Geotech J

Table 1 Comparison of PGA values obtained from different software packages for different damping ratios [20]
Software PGA (m/s2)
MCE DBE
15% Damping 10% Damping 5% Damping 15% Damping 10% Damping 5% Damping

Spec3 2.78 3.11 3.29 1.39 1.55 1.64

Wavgen 2.97 3.23 3.50 1.49 1.62 1.75
SeismoMatch 2.73 2.77 2.97 1.65 1.66 1.76

Fig. 7 Actual 1999 Chamoli

earthquake acceleration-time
history and response spectra
compatible time history of MCE
(obtained by SeismoMatch for
10% damping) [20]

Fig. 8 Geometry of soil–tunnel

system

Table 2 Variation of elastic modulus of Delhi silt with depth [23] boundary. Actual 1999 Chamoli earthquake and all
response spectra compatible time histories corresponding
Depth (m) Thickness (m) Elastic modulus (kPa)
to 10% damping have been considered for analysis. The
0–10 10 7500 seismic response of soil–tunnel system has been plotted in
10–20 10 15,000 the form of different time histories, and the response has
20–35 15 30,000 been presented here in the form of induced acceleration,
35–50 15 40,000 horizontal displacements in the system, and forces in RC
50–60 10 50,000 liners.

both vertical boundaries. No boundary condition was

applied along the horizontal base boundary, and the
earthquake was applied at the base of the model along this

123
Indian Geotech J

Displacement Response stresses are compressive in nature. It can be concluded that

for acceleration-time history considering MCE target
Values of maximum horizontal displacements induced in response spectra, Spec3 produces higher values of forces
soil–tunnel system during the earthquake are presented in and moments in RC liners as compared to those given by
Table 3 for different response spectra compatible time Wavgen, which in turn are marginally higher than those
histories as well as for the actual earthquake. These values given by SeismoMatch. This observation is also valid for
are presented for ground surface, crown, invert, and the response spectra time history considering DBE. DBE-
springing points of tunnel (Fig. 8). based response spectra give values of forces and moments
It can be noticed from Table 3 that Spec3 and Wavgen which are on lower side as compared to those given by
algorithms significantly overestimate the horizontal dis- MCE-based response spectra. Moreover, DBE-based target
placements, whereas SeismoMatch predicts horizontal response spectra produce forces and moments in liners
displacements which are comparable with those for both which are almost same for all three time histories.
MCE and DBE and are also comparable with those
obtained for the horizontal component of actual 1999 Induced Acceleration
Chamoli earthquake. It can therefore be concluded that
time history generated by SeismoMatch gives more reliable Time histories of output acceleration plotted at the crown
displacement response compared to those given by Spec3 of tunnel are presented in Fig. 9 for different applied
and Wavgen algorithms. Similarly, values of maximum compatible time histories and for the maximum considered
vertical displacements induced in soil–tunnel system dur- earthquake (MCE) target spectra. It can be observed from
ing the earthquake are presented in Table 4 for different Fig. 9 that induced values of acceleration at tunnel crown
response spectra compatible time histories as well as the are, respectively, 1.64 m/s2, 1.43 m/s2, and 1.50 m/s2, for
actual earthquake. It can be seen that vertical displace- all the three compatible time histories. Figure 10 represents
ments at all critical locations show a very good mutual the induced acceleration time history at ground surface for
comparison for all time histories of earthquake. It can as the actual 1999 Chamoli earthquake. It shows that maxi-
well be noticed that vertical displacement is maximum at mum induced acceleration of 2.26 m/s2 at the crown point
the crown point of tunnel. Similarly, modified time histo- is greater than the corresponding values obtained for the
ries were also generated considering the target response response spectra time histories. It can be therefore inferred
spectra of DBE by Spec3, Wavgen, and SeismoMatch that actual earthquake may overestimate the response of the
softwares, respectively. The phase and time interval of system and therefore conversion of actual time history into
modified time history are same as the actual earthquake. response spectra time history is very essential.

Forces in RC Liners Remarks

Values of axial thrust (T), shear forces (V), and bending Based on the above analysis, following concluding remarks
moment (M) induced in RC liners are also presented in can be made:
Table 5 for different time histories. As per the Plaxis sign If earthquake data are not available for a particular site,
convention, tension is positive and negative sign shows that then earthquake record of another similar site could be

Table 3 Maximum horizontal displacements of soil–tunnel system during earthquake for different response spectra compatible time histories
[20]
Location Maximum horizontal displacement Ux (mm)
1999 Chamoli EQ, actual Response spectra compatible time history Response spectra compatible time history
(MCE) (DBE)
Spec3 Wavgen SeismoMatch Spec3 Wavgen SeismoMatch

Ground surface (A) 264 2975 1094 209 1490 532 205
Crown (B) 154 2973 1502 185 1480 535 174
Invert (C) 151 2973 1485 180 1480 536 165
Springing point (1) 119 2013 1133 149 1522 583 136
Springing 185 2059 1067 216 996 516 203
point (2)

123
 Indian Geotech J

Table 4 Maximum vertical displacements of soil–tunnel system during earthquake for different response spectra compatible time histories [20]
Location Maximum vertical displacement Uy (mm)
Actual 1999 Chamoli earthquake Response spectra compatible time history Response spectra compatible time history
(MCE) (DBE)
Spec3 Wavgen SeismoMatch Spec3 Wavgen SeismoMatch

Ground surface 35 35 31 31 30 29 29
Crown 77 79 76 76 76 75 74
Invert 50 48 49 49 48 49 49
Springing point (1) 20 18 18 18 16 16 15
Springing point (2) 14 20 16 15 16 15 14

Table 5 Maximum forces in RC liners during earthquake for different time histories [20]
Max forces in RC Static 1999 Chamoli earthquake, Response spectra compatible time Response spectra compatible time
liners analysis actual history (MCE) history (DBE)
Spec 3 Wavgen Seismo Spec 3 Wavgen Seismo
Match Match

Tmax (kN/m) - 534.98 - 627.99 - 610.9 - 598.62 - 595.84 - 562.4 - 559.76 - 561.27
Vmax (kN/m) 170.20 - 211.54 210.80 200.30 196.45 182.30 179.70 172.07
Mmax (kN-m/m) - 256.83 319.69 327.93 307.56 302.04 273.53 270.34 270.34

Fig. 9 Horizontal acceleration

time history at crown of the
tunnel for different time
histories (TH), considering
MCE [20]

taken for analysis. Actual earthquake for another site may compatible time history produced results which are quite
overestimate or underestimate the response at the proposed similar when compared with those for the actual
site, and therefore, conversion of actual time history into earthquake.
response spectra compatible time history is essential.
All three response spectra compatible time histories
yield comparable response in terms of induced accelera- Significance of Three-Dimensional Analysis
tion, vertical displacement, and forces in RC liners. How-
ever, Spec3 and Wavgen highly overestimate the horizontal FLAC 3D, an explicit finite difference-based program, was
displacements as compared to those predicted by Seis- used for understanding three-dimensional behavior of
moMatch. SeismoMatch-based response spectra shallow tunnels under seismic loading [25]. The study

123
Indian Geotech J

Fig. 10 Horizontal acceleration

time history at crown of the
tunnel for actual 1999 Chamoli
earthquake [20]

presented three-dimensional nonlinear behavior of a tunnel dimensional seismic behavior of underground tunnels
excavated in soft soil and subjected to seismic loading. The [20, 27–33, 34].
originality of research was about the inclusion of tunnel
excavation sequence in the definition of the stress state of Delhi Metro Tunnel Problem
soil before the earthquake. For that, tunnel excavation was
effected first for generating the initial state of stress in soil, In this study, three-dimensional finite element analysis of
and subsequently, the model was subjected to a seismic one of the life line structures, i.e., Delhi Metro Under-
excitation. The response parameters included the internal ground tunnels, has been carried out considering response
forces induced in tunnel lining (axial thrust, bending spectra compatible time history of 1999 Chamoli earth-
moment, and shear force) and the settlement at the ground quake. Artificial time histories of all three components of
surface. It was recommended that elastic analysis is not earthquake, i.e., horizontal/transverse (T), longitudinal (L),
adequate to determine the earthquake induced response of and vertical (V), have been generated and used to carry out
soil–tunnel system. The influence of tunnel excavation both linear and nonlinear dynamic analyses and the com-
should be integrated in the model which would lead to a parison made. Attempt has also been made to compare the
more realistic estimation of the seismically induced thrust dynamic response obtained using three different boundary
in the tunnel lining and also the surface settlement. conditions, namely elementary boundary, absorbent
An attempt was also made to understand the seismic boundary, and the free-field boundary. Moreover, com-
response of tunnels in longitudinal (axial) direction by parison has also been made with the 2-D plane-strain
carrying out 3-D numerical analysis under an asynchronous response. All the data of DMRC tunnel remain the same as
wave input [26]. 1D time-domain approach was applied to that presented in Fig. 10 and Table 2; the only difference is
calculate the free-field motion. Based on the numerical that the depth of overburden here is 12.0 m and damping
approach for the external source wave motion problems, ratio of Delhi silt considered is 15%. No water table was
the wave input was determined by applying equivalent encountered during the tunnel excavation.
nodal force at the truncated boundary, and its accuracy and The extent of model, after carrying out sensitivity
validity were proved by numerical examples. This wave analysis, has been taken as 200 m 9 50 m 9 60 m in X-,
input method was directly applied to a 3-D soil–tunnel Y-, and Z-directions, respectively. Ten-noded tetrahedral
structure interaction model to simulate the longitudinal elements with an average size of 16.16 m (less than the
response of tunnels subjected to asynchronous earthquakes. wavelength of incoming wave, k = 27.65 m) were con-
The results proved the validity of modeling the seismic sidered for modeling of soil domain. Elastic behavior of
response of a tunnel with an infinite length as long as the soil was considered. Segmental RC liners of tunnel were
computational model was relatively long in the longitudi- simulated using 6-noded triangular plate bending elements.
nal direction. The influence of the angle of incidence and For RC liners, elastic behavior was considered. No-slip
the incident direction on the longitudinal seismic response condition was assumed between tunnel and the surrounding
of the tunnel was considered. Several preliminary conclu- soil medium. For obtaining static response, boundary dis-
sions were drawn regarding further study and research. placement conditions were applied along the four vertical
Many other authors have also studied the three- surfaces and one bottom surface. Displacements of nodes

123
 Indian Geotech J

along four vertical planes were restrained in the direction Displacements in Soil–Tunnel System
normal to the planes, whereas displacements of all nodes
lying on the bottom plane were completely restrained. For Maximum dynamic displacements in soil–tunnel system
dynamic analysis, viscous absorbent boundary [24] was due to application of different components of earthquakes
used to represent the displacement condition along both are presented in Table 6. It can be seen that maxima values
vertical planes. of horizontal, vertical, and longitudinal displacements, both
in soil medium and RC liners of tunnel, occur in respective
Earthquake Loading directions of T, V, L components of earthquake, respec-
tively. For T component, maxima values of horizontal
All three components, i.e., horizontal (T), vertical (V), and displacement (Ux) predicted are of the order of 23.53 mm
longitudinal (L), of the 1999 Chamoli earthquake of the in soil medium and 20.50 mm in the tunnel. For V com-
lower Himalaya have been considered for seismic analysis. ponent, maximum values of vertical displacement (Uz)
Response spectra compatible modified time histories of this have been found to be of the order 207.00 mm in soil
earthquake were generated for all T, V, and L components medium and 163.04 mm in the tunnel. Similarly, for L
[20, 34] which are presented in Fig. 11. These histories component, maximum values of longitudinal displacement
were applied in turn at the nodes on the bottom plane of the (Uy) are 193.5 mm in soil medium and 27.40 mm in tunnel.
model in X-, Z-, and Y-directions, respectively. The com-
ponents were applied separately (one by one) and not Forces in RC Liners
simultaneously. These components have PGA values of
2.713 m/s2, 2.044 m/s2, and 3.032 m/s2, respectively. Table 7 gives the values of residual forces in RC liners
after the earthquake for different components of earth-
Deformed Mesh quake. From Table 7, it can be seen that L component of
earthquake has produced significantly higher values of
Figures 12, 13, and 14 show, respectively, the deformed forces in RC liners than due to other components of
mesh due to T, V, and L components of 1999 Chamoli earthquake. However, it has been found that the corre-
earthquake. Values of resultant displacement are of the sponding axial stress mobilized in the liners is less than
order of 23.53 mm for T component, 207.00 mm for V 1.0 MPa and hence much less than the permissible tensile
component, and 193.5 mm for L component of the earth- strength of concrete.
quake. Therefore, it may be noted that the maximum value
of resulting displacement has been obtained due to the Induced Acceleration
vertical component (V) rather than other two horizontal
components of earthquake, namely T and L. Time histories of induced acceleration are plotted in
Figs. 15, 16 and 17. For T component of earthquake, it can
be seen from Fig. 15 that it is the ground surface that
experiences maximum horizontal acceleration of 1.23 m/s2
equivalent to 0.125 g or 45.3% of the maximum applied

Fig. 11 Response spectra

compatible time history for
horizontal (Ax), vertical (Az),
and longitudinal (Ay)
components of 1999 Chamoli
earthquake [20, 34]

123
Indian Geotech J

Fig. 12 Deformed mesh due to

T component of earthquake
[20, 34]

Fig. 13 Deformed mesh due to

V component of earthquake
[20, 34]

horizontal acceleration, whereas acceleration level experi- component of earthquake (Fig. 17), it can be noticed that it
enced at the tunnel crown is 0.78 m/s2, at tunnel invert, it is is again the ground surface that experiences maximum
0.95 m/s2, and that at the springing points, it is 0.82 m/s2. longitudinal acceleration of 0.32 m/s2 equivalent to
For the V component of earthquake, it can be observed in 0.032 g or 10.5% of the maximum applied horizontal
Fig. 16 that it is the ground surface again that experiences acceleration in Y or L direction, whereas acceleration level
maximum horizontal acceleration of 1.65 m/s2 equivalent experienced at the tunnel crown is 0.23 m/s2, at tunnel
to 0.168 g or 80.7% of the maximum applied vertical invert, it is 0.26 m/s2 and that at the springing points, it is
acceleration, whereas acceleration level experienced at the 0.28 m/s2.
tunnel crown is 1.21 m/s2, at tunnel invert, it is 1.04 m/s2
and that at the springing points, it is 1.12 m/s2. For the L

123
 Indian Geotech J

Fig. 14 Deformed mesh due to

L component of earthquake
[20, 34]

Table 6 Maximum dynamic displacements in soil–tunnel system after the earthquake [20, 34]
U (mm) Soil medium Tunnel
T comp. of EQ V comp. of EQ L comp. of EQ T comp. of EQ V comp. of EQ L comp. of EQ

Ux 23.53 66.72 26.66 20.50 8.626 1.306

Uy 4.74 78.31 193.50 0.024 0.82 27.40
Uz 8.23 207.00 54.76 0.25 163.04 7.84

Table 7 Forces in RC liners at the end of earthquake [20, 34]

Forces dF
T component of earthquake V component of earthquake L component of earthquake

N1 (kN) - 1.45 51.77 5004.48

N2 (kN) 0.30 6.60 694.545
Q12 (kN) 4.44 11.95 570.420
Q23 (kN) 0.12 - 2.40 33.320
Q13 (kN) 0.66 - 3.98 9.920
M11 (kN-m) - 0.016 - 0.081 27.710
M22 (kN-m) 0.12 - 3.20 21.170
M12 (kN-m) 0.16 0.218 11.322
N1 is the axial force in Y-direction, N2 is the force in transverse direction perpendicular to tunnel axis, (Y), Q12 is the in-plane shear force in YZ
plane, Q13 is the shear force in XY plane, Q23 is the shear force in YZ plane, M11 is bending moment due to bending over the Z-axis (around Z-
axis), M22 is bending moment due to bending over the X-axis (around X-axis), and M12 is torsional moment according to transverse shear force.

Comparison of 3D Analysis with 2D Plane-Strain strain analysis. The comparison has been made both for T
Analysis and V components of earthquake in terms of horizontal
displacement and vertical displacement (Table 8). Three-
3D dynamic response of metro underground metro tunnel dimensional analysis leads to overall reduction in the
has been compared with response predicted from 2D plane- horizontal and vertical displacements.

123
Indian Geotech J

Fig. 15 Time history of

horizontal acceleration at the
different locations (T
component of earthquake) [20]

Fig. 16 Time history of vertical

acceleration at different
locations (V component of
earthquake) [20]

Fig. 17 Time history of

longitudinal acceleration at
different locations (L
component of earthquake) [20]

Remarks displacements were experienced in soil–tunnel system

when the system was subjected to T, V, and L components
All the displacements in soil–tunnel system were found to of earthquake, respectively. Residual forces in RC liners
be maximum at the ground surface during the earthquake. were found to increase only in case when L component of
Higher values of horizontal, vertical, and longitudinal earthquake was applied. In the overall view, 3D analysis

123
 Indian Geotech J

Table 8 Maximum displacements, U (mm) during the earthquake [20]

Points Ux due to T component of earthquake Uz due to V component of earthquake
2D plane-strain analysis 3D analysis 2D plane-strain analysis 3D analysis

G.S 329 201.36 259 227.47

Crown 311 186.77 253 220.45
Invert 301 184.52 251 219.97
S.P. (1) 306 183.7 252 220.00
S.P. (2) 306 183.5 252 222.00

yields much subdued dynamic response as against the increase in overburden depth, (2) liquefaction potential of
response obtained in 2D plane-strain analysis. soils was dependent on the overburden depth of tunnel and
not on its diameter and thickness, (3) a high uplift is
mobilized in the tunnel and also the surrounding ground,
Effect of Liquefaction on Lifeline Structures and (4) forces in tunnels were also found to increase sig-
nificantly due to earthquake. Moreover, different earth-
The factors which affect the liquefaction phenomenon quake loading conditions influence the pore water pressure
[35, 36] include: (1) type of soil, (2) shape of soil particles, distribution and subsequently alter the stresses in RC liners.
(3) grain size distribution, (4) permeability of soil, (5) Increasing the loading frequency causes reduction of forces
depth of ground water table, (6) historical environment, (7) in the tunnel lining and pore water pressure, although as the
age of soil, (8) confining pressure, (9) relative density, (10) loading amplitude decreases, stresses in liners and exces-
natural soil deposits in water bodies, (11) location of sive pore water tend to reduce; v) deeper tunnels were
drainage and dimensions of deposit, (12) trapped air, and found to be less vulnerable to liquefaction than shallow
(13) presence of seismic waves. Liquefaction can damage tunnels. The acceleration response at different depths was
shallow life line structures if they are constructed in strata found to attenuate significantly due to occurrence of soil
like poorly saturated soft sandy soils. Life line structures liquefaction; vi) the increase in amplitude of vertical
located below the ground water table in liquefiable soils acceleration increased the liquefaction area, ground surface
can experience: (1) increases in lateral pressure, (2) deformation, and uplift of the tunnel structure, but had
reduction in lateral passive resistance, (3) sinking or small influence on internal forces in the tunnel structure.
flotation in liquefied soil, (4) lateral displacements, (5) The effect of liquefaction on seismic response of Delhi
permanent settlements, and (6) tension and compression Metro Underground tunnels was undertaken by [20, 44]
failure after the dissipation of pore water pressure and the and found that tunnels can suffer damage only if the water
consequent consolidation of soil [37]. table in Delhi area rises to the level of the crown of tunnels,
The seismic response of large underground tunnels which will not generally happen because the water
constructed in liquefiable soils due to horizontal and ver- table will have to rise by at least 10 m. Various life line
tical components of earthquake was studied by using either structures in India are at shallow depth and are susceptible
the finite difference method or the finite element method by to damage in the event of any major earthquake. So it is
many investigators [38–42, 43]. It was concluded on basis important to understand the seismic response of these
of these studies that (1) safety of these underground structures excavated in soils and investigate if they expe-
structures against liquefaction damage improves with rience any damage due to liquefaction. In this section, an

Table 9 Properties of soil surrounding the tunnel [23]

Depth E (MPa) Nm Strength parameters

0–10 7.54 5 Cohesion, c/ = 0

10–20 18.94 27 Friction angle, u/ = 35°
20–30 30.34 49 Dilation angle, w/ = 5°
30–40 35.86 59
40–50 53.14 92
50–60 64.54 113

123
Indian Geotech J

Table 10 Properties of soil for UBC3D-PLM model [20, 23, 44] Table 12 Properties of soil for hardening soil (HS) model
[20, 23, 44]
Parameters Values
Properties Symbol Values
Peak friction angle, up (in deg.) 35
Constant volume friction angle, ucv (in deg.) 29.1 Cohesion c/ 0
/
Elastic shear modulus index, me 0.5 Friction angle u 35°
Elastic bulk modulus index, ne 0.5 Dilation angle w/ 5°
Plastic shear modulus index, np 0.5 Poisson’s ratio for unloading m 0.2
Failure ratio, Rf 0.7 Reference pressure Pref 100
Atmospheric pressure, PA (kPa) 100 K0 value for normal consolidation K0nr 0.426
Tension cutoff, rt (kPa) 0 Failure ratio Rf 0.7
Densification factor, fachard 0.45 Power for stress level dependency of stiffness m 1
Post liquefaction factor, facpost 0.2

(HS) model was considered for modeling the soil, whereas

attempt has been made to carry out three-dimensional UBC3D-PLM model was used to model the soil below the
liquefaction analysis of one of the life line structures, i.e., water table. As liquefaction occurs only in saturated soils,
Delhi Metro Underground tunnels using Plaxis 3D soft- therefore there is no need to use UBC3D-PLM model in
ware. Parametric study was also undertaken by considering unsaturated soil. The UBC3D-PLM model was first
different parameters like seismic directions, depth of water developed by [45]. This model is an extension of the two-
table, and peak ground acceleration (PGA). dimensional UBC-SAND model developed at University of
British Colombia.
Analysis of DMRC Tunnel for Liquefaction Some important state variables used in UBC3D-PLM
model are tabulated in Table 10. Pore pressure ratio, PPR,
All other parameters remain unchanged; the only differ- can show the current status during the calculation proce-
ence is that the depth of overburden above the tunnel is dure, and it is given by Eq. (1).
10.60 m and the depth of water table is 17.0 m. In situ unit =
Pi  P=c
weight, cbulk and saturated unit weight, csat of Delhi silt PPR ¼ =
ð1Þ
were found to be 18 kN/m3 and 20 kN/m3, respectively. Pi
Damping of soil was taken as 15% for the analysis, and the where P/i is the initial effective mean stress and P/c is the
analysis has been carried out for an undrained condition. current effective mean stress. Maximum pore pressure ratio
For this study, engineering properties of alluvium have (PPRm) can reveal if the soil liquefies even once during the
been adopted from [23] and are summarized in Table 9. earthquake. The state variable, ru, gives similar information
as PPR but instead of the effective mean stress, the vertical
Numerical Modelling for Liquefaction effective stress is used as shown in Eq. (2), [45].
= =
In order to study the liquefaction response of metro rvertical;i  rvertical;c
ru ¼ =
ð2Þ
underground tunnels, the above three-dimensional finite rvertical;i
element model was reconsidered for analysis using Plaxis
3D software. Two different material models were taken for Parameters defined in UBC3D-PLM model are
modeling the soil. Above the water table hardening soil calculated in Tables 10 and 11. Some parameters are

Table 11 Modulus number and SPT values of soil for UBC3D-PLM model [20, 23, 44]
Depth Nm Corrected SPT value, (N1)60 Elastic shear modulus, KGe Elastic bulk modulus, KBe Plastic shear modulus, KGP

0–10 5 6 788.2 438.2 185.1

10–20 27 16 1092.6 607.5 939.1
20–30 49 24 1250.5 695.3 2260.9
30–40 59 27 1300.57 723.1 2944.3
40–50 92 37 1444.4 803.1 6032.3
50–60 113 42 1506.7 837.7 8073.55

123
 Indian Geotech J

Table 13 Moduli values (MPa) for hardening soil (HS) model [20, 23, 44]
Depth Initial Tangent modulus, Secant stiffness in SDT Tangent stiffness for primary Unloading/reloading stiffness
modulus, E50 (= 0.65 Ei) test, E50ref (= E50) oedometer loading Eurref (= 3 E50ref) Eoedref (= E50ref / 1.25)
Ei

0–10 7.54 4.9 4.9 14.7 3.92

10–20 18.94 12.31 12.31 36.93 9.85
20–30 30.34 19.72 19.72 59.16 15.78
30–40 35.86 23.31 23.31 69.93 18.65
40–50 53.14 34.54 34.54 103.62 27.63
50–60 64.54 41.95 41.95 125.85 33.56

Fig. 18 Total displacement in

soil–tunnel system at end of T
component of 1999 Chamoli
earthquake [20, 44]

Table 14 Static and dynamic displacements is soil–tunnel system (T component of earthquake) [20, 44]
U Soil medium RC liners
Static displacement Dynamic displacement Static displacement Dynamic displacement

Ux (mm) 3.42 183.8 3.584 10.90

Uy (mm) 1.153 74.6 0.028 1.98
Uz (mm) 5.405 335.4 5.131 54.9

kept constant for all the layers of soil, and values of these earthquake loading were kept same as in Sect. 5.0 (3D
parameters are stated in Table 10, whereas modulus analysis).
numbers and corrected SPT values vary with depth and The response obtained after dynamic analysis of DMRC
are stated in Table 11. Properties of hardening soil material tunnels, when subjected to T component (X-direction) of
model have been calculated and are presented in Tables 12 earthquake, is presented below for discussion.
and 13. Some properties are kept constant with depth of
soil and are presented in Table 12, whereas the stiffness Displacements
varies with depth of soil and values of soil moduli with
depth are stated in Table 13. Boundary conditions and Figure 18 shows the contours of total displacement in soil–
tunnel system. Maximum displacement in soil medium is

123
Indian Geotech J

Fig. 19 Maximum pore

pressure ratio in soil during the
earthquake [20, 44]

Table 15 Effect of seismic directions on liquefaction [20, 44]

Parameters T comp. of EQ in X-direction V comp. of EQ in Z-direction L comp. of EQ in Y-direction

PPRm 0.55 0.65 1.244

PPR 0.40 0.42 1.228
rum 0.65 0.67 1.34
ru 0.60 0.48 1.34

Fig. 20 Pore pressure ratio

during L component of 1999
Chamoli earthquake [20, 44]

of order of 449.2 mm. The maximum total displacement in Liquefaction Susceptibility

the tunnel at the end of T component of earthquake
(residual) was found to be 55.37 mm, as shown in Fig. 18. Liquefaction susceptibility in soil surrounding the tunnel
Values of static and dynamic displacements in soil mass can be checked in terms of pore pressure ratio (PPR) and
and RC liners of tunnel are presented in Table 14. It can be the state variable (ru). It may be noted that water table ex-
seen that transverse (T) component of earthquake causes ists at 17.0 m below the ground surface. Distribution of
significant increase in horizontal (X-direction) and vertical maximum pore pressure ratio during the earthquake is
(Z-direction) displacements in soil medium as well as in presented in Fig. 19.
RC liners. However, dynamic displacement in longitudinal It can be seen that maximum value of pore pressure ratio
direction, i.e., in Y-direction, is much smaller in soil mass occurs below the tunnel invert. Pore pressure ratio (PPRm)
and is negligible in RC liners. was found to be 0.4 for various points below the invert.
Overall maximum value of pore pressure ratio is 0.55

123
 Indian Geotech J

Table 16 Effect of position of water table on liquefaction [20, 44]

Parameters Water table depth, h (m)
h = 0 m (G.S.) h = 10 m (above the crown) h = 17 m (below the invert) h = 30 m h = 40 m h = 50 m

PPRm 1.235 0.815 0.55 0.451 0.451 0.45

PPR 1.232 0.845 0.40 0.225 0.125 0.125
rum 1.199 0.852 0.65 0.40 0.36 0.25
ru 1.199 0.852 0.60 0.32 0.24 0.18

which has been found to occur at points just below the Effect of Depth of Water Table
water table (as red color shown in Fig. 19). Pore pressure
ratio (PPR) after the earthquake was found to be of the Dynamic analysis has therefore been carried out for hori-
order of 0.4 in layer 3, i.e., between depths of 17–20 m. It zontal, T component (X-direction) of 1999 Chamoli
was found to be less than 1.0 for both during the earthquake earthquake by considering different depths of water table.
and after the earthquake conditions. Therefore, it can be Values of liquefaction parameters obtained for different
concluded that soil medium surrounding the tunnel is safe depths of water table ranging between 0.0 and 50 m are
against liquefaction. Similarly, susceptibility of liquefac- presented in Table 16. Initially, when actual depth of water
tion can be checked in terms of state variable, ru. It can be table was 17 m, the soil medium surrounding the tunnel
found that maximum value of state variable (ru) is of the was found to be safe against liquefaction. When depth of
order of about 0.65 during the earthquake and it is 0.60 at water table rises to a level just above the crown of tunnel
the end of earthquake. Therefore, this parameter also sug- (h = 10 m), soil medium surrounding the tunnel was not
gests that soil surrounding the tunnel is safe against the found to liquefy. It can also be observed that liquefaction
liquefaction. occurs in soil mass when water table rises to the ground
surface. In that case, both pore pressure ratio and state
Parametric study variable (ru) assume values greater than unity. Therefore, it
can be concluded that in present condition (i.e., when water
Effect of Seismic Directions on Liquefaction table depth, h = 17 m), Delhi metro tunnels are safe
against liquefaction in the event when seismic waves
Values of liquefaction parameters have been obtained for propagate in the direction of T component of 1999 Chamoli
all three components of earthquake and are presented in earthquake. However, liquefaction can occur if water
Table 15, from which it can be noticed that (1) values of table rises up close to the ground surface.
liquefaction parameters are less than unity when seismic
waves propagate either in horizontal, X-direction or in Effect of PGA
vertical, Z-direction, and (2) values of both pore pressure
ratio and the state parameter (ru) are greater than unity In this section, the influence of peak ground acceleration
when L component (Y-direction) of earthquake was applied (PGA) on liquefaction has been discussed. For this, depth
along the length of tunnel. of water table was retained at 17 m below the ground
Therefore, liquefaction can occur due to the L compo- surface. T component of Chamoli earthquake (PGA =
nent of 1999 Chamoli earthquake when seismic waves 2.71 m/s2) was taken for analysis. The PGA of this
propagate in the direction parallel to the alignment of earthquake was scaled up to 1.5 m/s2, 3.6 m/s2, 4.0 m/s2,
tunnel. It can as well be seen from Fig. 20 that liquefaction and 5.0 m/s2. Values of liquefaction parameters thus
occurs in layer 3, i.e., at depth between 17 and 20.0 m. obtained are presented in Table 17. It has been found that

Table 17 Effect of PGA on liquefaction [20, 44]

Parameters PGA = 1.5 PGA = 2.71 PGA = 3.6 PGA = 4.0 PGA = 5.0
m/s2 m/s2 m/s2 m/s2 m/s2

PPRm 0.52 0.55 0.61 0.679 1.018

PPR 0.34 0.40 0.50 0.55 0.955
rum 0.36 0.65 0.70 0.74 1.039
ru 0.36 0.60 0.69 0.74 1.03

123
Indian Geotech J

both pore pressure ratio and state variable (ru) increase a seismic event when seismic waves travel in a
with increase in PGA. But this Delhi site was found to be direction parallel or nearly parallel to the alignment
safe against liquefaction during this earthquake even when of tunnels.
PGA was scaled to 4.0 m/s2. Soil mass enters the liquefied 7. In the present condition when water table depth is at
state only when PGA is scaled beyond up to 5.0 m/s2. 17m below the ground surface, Delhi metro tunnels are
safe against liquefaction as far as T component of 1999
Remarks Chamoli earthquake is concerned. However, there can
be a cause of concern only if water table rises above to
Soil medium surrounding the tunnel is safe against lique- the ground surface.
faction for T (horizontal) and V (vertical) components of 8. Delhi metro tunnels are also safe against liquefaction
earthquake but liquefaction can occur in case of a seismic for 1999 Chamoli earthquake even when PGA of T
event when seismic waves travel in a direction parallel or component of earthquake (in x-sec. plane) is scaled up
nearly parallel to the alignment of tunnels. In the present to 4.0 m/s2. However, it was found that soil enters into
condition, when water table depth is at 17 m below the liquefied state when PGA is scaled up to 5.0 m/s2
ground surface, Delhi metro tunnels are safe against the (0.5096g).
liquefaction as far as horizontal, T component of Chamoli
earthquake is concerned but may become unsafe in case
when water table rises above to the ground surface. Delhi Funding Necessary finding for this work was received from The
metro tunnels are also safe against liquefaction for this Ministry of Human Resources Development (MHRD), New Delhi,
earthquake even when PGA of horizontal, T component of India.
earthquake is scaled up to 4.0 m/s2. However, it was found
Compliance with Ethical Standards
that soil enters into liquefied state when PGA is scaled up
to 5.0 m/s2 (i.e., 0.50968 g). Conflict of interest The authors declare that they have no known
conflicting financial or any other interests.

Summary and Conclusions

References
1. If earthquake data are not available for a particular site,
1. Hashash YA, Hook J, Schmidt B, Chiangyao J (2001) Seismic
then earthquake record of another similar site should design and analysis of underground structures. J Tunnel Undergr
be considered for analysis. Actual earthquake for Space Technol 16(4):247–293
another site may overestimate or underestimate the 2. Dowding CH, Rozen A (1978) Damage to rock tunnels from
response at the proposed site, and therefore, conversion earthquake shaking. J Geotech Eng 2:175–191
3. Kumar A (2006) Software for generation of spectrum compatible
of actual time history into response spectra compatible time history having same phase as of a given time history. In:
time history is very essential. Proceedings 8th U.S. National conference on earthquake engi-
2. Presence of tunnels modifies the response at the ground neering, San Francisco, California, USA, April 18–22, Paper No.
surface as compared to that for a free-field condition. 172
4. Abrahamson NA (2010) Nonstationary spectral matching pro-
3. Tunnels with smaller H/D ratio or with lesser depth of gram RSPMATCH. User’s manual (1992)
overburden are more prone to damage. Forces in RC 5. Gasparini DA, Vanmarcke EH (1976) Simulated earthquake
liners and settlements in soil–tunnel system are higher motion compatible with prescribed response spectra. Mas-
in shallow tunnels than in deep tunnels. sachusetts Institute of Technology, Department of Civil Engi-
neering, Constructed Facilities Division, Springfield, VA:
4. Three-dimensional analysis yields much subdued Distributed by National Technical Information Service, U.S.
dynamic response as against the response obtained in Dept. of Commerce
2D plane-strain analysis. 3D analysis leads to overall 6. Silva WJ, Lee K (1987) WES RASCAL Code for synthesizing
reduction in horizontal and vertical displacements and earthquake ground motions. Miscellaneous paper S-73-1. Report
24, US Army Core of Engineers, Waterways Experiment Station,
induced acceleration. However, residual forces in RC Vicksburg, Missisippi, p 73
liners after the earthquake were found to increase in 3D 7. Lilhanand K, Tseng WS (1988) Development and application of
analysis, the increase due to V component of earth- realistic earthquake time histories comparable with multiple
quake being quite significant. damping design spectra. In: Proceedings 9th world conference on
earthquake engineering, Tokyo-Kyoto, Japan, vol 2
5. Safety of underground structures against liquefaction 8. Bolt BA, Gregor NJ (1993) Synthesized strong ground motions
damage improves with increase in overburden depth. for the seismic condition assessment of the eastern portion of the
6. Soil medium surrounding the tunnels is safe against San Francisco Bay Bridge. Report UCB/EERC-93/12, Earth-
liquefaction for transverse and vertical components of quake Engineering Research Center, University of California,
Berkeley, CA
earthquake. However, liquefaction can occur in case of

123
 Indian Geotech J

9. Gupta ID, Joshi RG (1993) On synthesizing response spectrum 29. Yu H, Yuan Y, Bobet A (2017) Seismic analysis of long tunnels:
compatible accelerograms. Proc Eur Conf Earthq Eng 7(2):25–33 a review of simplified and unified methods. J Undergr Space
10. Shrikhande M, Gupta VK (1996) On generating ensemble of 2:73–87
design spectrum-compatible accelerograms. J Eur Earthq Eng 30. Yu H, Yuan Y, Chen Z, Yu G, Gu Y (2009) Full 3D numerical
3:49–56 simulation method and its application to seismic response anal-
11. Conte JP, Peng BF (1997) Fully nonstationary analytical earth- ysis of water-conveyance tunnel. J Comput Struct Eng
quake ground motion model. J Eng Mech ASCE 123(1):15–24 19:349–358
12. Shrikhande M, Gupta VK (2011) On the characterization of the 31. Chen Z, Yu H, Yuan Y (2014) Full 3D seismic analysis of a long-
phase spectrum for strong motion synthesis. J Earthq Eng distance water conveyance tunnel. J Struct Infrastruct Eng
5(4):465–482 10(1):128–140
13. Atik LA, Abrahamson N (2010) An improved method for non- 32. Ichimura T, Tanaka S, Hori M, Yamamoto Y, Dobashi H, Osada
stationary spectral matching. J Earthq Spectra 26(3):601–617 M, Ohbo N, Yamada T (2016) Full three-dimensional seismic
14. Singh M, Viladkar MN, Samadhiya NK (2017) Seismic response response analysis of underground structures with large complex
of Delhi metro underground tunnels. Int J Geotech Eng cross sections and two-step analysis method for reducing the
11(2):175–185 computational costs. J Earthq Tsunami 10(5):1640016
15. Trifunac MD, Udwadia FE, Brady AG (1971) High frequency 33. Chun LD, Qi LV, Li DX, Chao M (2015) Three dimensional
errors and instrument corrections of strong-motion accelerogram. seismic response analysis of tunnels and surrounding rock.
Earthquake Energy Res. Lab. EERL 71-05, CALTECH., Pasa- J Disaster Prev Mitig Eng 02
dena, California 34. Singh M, Viladkar MN, Samadhiya NK (2018) Three dimen-
16. Wong HL, Trifunac MD (1979) Generation of artificial strong sional seismic analysis of metro underground tunnels. In: Pro-
motion accelerograms. J Earthq Eng Struct Dyn 7:509–527 ceedings international conference advances in construction
17. Basu B, Gupta VK (1999) Wavelet-based analysis of non-sta- materials and structures, Department of Civil Engineering, IIT
tionary response of a slipping foundation. J Sound Vib Roorkee
222(4):547–563 35. Kramer SL (1996) Geotechnical earthquake engineering. Pearson
18. Fahjan Y, Ozdemir Z (2008) Scaling of earthquake accelero- Education Inc. and Dorling Kindersley Pub. Inc., London
grams for non-linear dynamic analyses to match the earthquake 36. Prakash S (1981) Soil dynamics. McGraw Hill, New York.
design spectra. In: 14th world conference on earthquake engi- Reprinted SP Foundation, Rolla, MO, USA
neering, Beijing, China, pp 12–17 37. Hashash YA, Hook J, Schmidt B, Chiangyao J (2005) Seismic
19. Katsanos EI, Sextos A, Manolis GD (2010) Selection of earth- design and analysis of underground structures. J Tunnel Undergr
quake ground motion records: a state-of-art review from a Space Technol 20(5):435–441
structural engineering perspective. J Soil Dyn Earthq Eng 38. Azadi M, Mohammad SM (2010) Analyses of the effect of
30(4):157–169 seismic behavior of shallow tunnels in liquefiable grounds.
20. Singh M (2018) Seismic response of shallow metro underground J Tunnel Undergr Space Technol 25(5):543–552
tunnels. Ph.D. thesis, Department of Civil Engineering, IIT 39. Azadi M (2011) The seismic behavior of urban tunnels in soft
Roorkee, India saturated soils. In: Proceedings 12th East Asia Pacific conference
21. IS: 1893 Part-1 (2002) Code of practice: criteria for earthquake on structural engineering & construction, procedia engineering,
resistant design of structures. Bureau of Indian Standards, (BIS), vol 14, pp 3069–3075
Manak Bhawan, New Delhi 40. Bao X, Xi Z, Ye G, Fu Y, Su D (2017) Numerical analysis of the
22. Sony S (2015) Static and dynamic response of Delhi metro tun- seismic behavior of a large metro subway tunnel in liquefiable
nels. Ph.D. thesis, Department of Civil Engineering, IIT Delhi, ground. J Tunnel Undergr Space Technol 66:91–106
India 41. Liu H, Song E (2005) Seismic response of large underground
23. Yadav HR (2005) Geotechnical evaluation of Delhi metro tun- structures in liquefiable soils subjected to horizontal and vertical
nels. Ph.D. Thesis, Department of Civil Engineering, IIT Delhi, earthquake excitations. J Comput Geotech 32(4):223–244
India 42. Salehzadeh H, Parchami ML (2014) Seismic analysis of twin
24. Lysmer J, Kuhlemeyer RL (1969) Finite dynamic model for tunnels in liquefiable soils. J Transp Res 2(39):125–158
infinite media. J Eng Mech 95:859–878 43. Unutmaz B (2014) 3D liquefaction assessment of soils sur-
25. Sliteen I, Mroueh H, Sadek M (2011) Three-dimensional mod- rounding circular tunnels. J Tunnel Undergr Space Technol
eling of the behavior of shallow tunnel under seismic loading. 40:85–94
Laboratory of Civil Engineering and GeoEnvironment (LGCgE), 44. Viladkar MN, Singh M, Samadhiya NK (2019) Liquefaction
University of Lille, 1-Polytech-Lille, France analysis of metro underground tunnels. In: 7th international
26. Li P, Song EX (2015) Three-dimensional numerical analysis for conference earthquake geotechnical engineering, Rome, Italy,
the longitudinal seismic response of tunnels under an asyn- June
chronous wave input. J Comput Geotech 63:229–243 45. Tsegaye A (2010) Plaxis liquefaction model. External report.
27. Stamos AA, Beskos DE (1995) Dynamic analysis of large 3-D Plaxis knowledge base, http://www.plaxis.nl
underground structures by the BEM. J Earthq Eng Struct Dyn
24(6):917–934 Publisher’s Note Springer Nature remains neutral with regard to
28. Stamos AA, Beskos DE (1996) 3-D seismic response analysis of jurisdictional claims in published maps and institutional affiliations.
long lined tunnels in half-space. J Soil Dyn Earthq Eng
15(2):111–118

123