Soil Dynamics and Earthquake Engineering 164 (2023) 107619

Contents lists available at ScienceDirect

Soil Dynamics and Earthquake Engineering

journal homepage: www.elsevier.com/locate/soildyn

Tunnel-soil-bridge seismic interaction on soft clay

Juan Manuel Mayoral *, Daniel De La Rosa, Mauricio Alcaraz, Enrique Barragan
Institute of Engineering, National Autonomous University of Mexico, Mexico

A R T I C L E I N F O A B S T R A C T

Keywords: Seismic tunnel-soil-bridge interaction in high plasticity soft clays can lead to significant ground motion vari­
Tunnel-bridge systems ability, affecting on-ground structures located next to or above underground facilities. This technical note pre­
Soil-structure interaction sents a numerical study aiming at establishing tunnel-bridge seismic interaction effects on soft clays. Initially, we
Ground motion incoherence
revised this interaction considering the impact that frequency content, intensity, and strong ground motion
duration have on the interplay between incoming seismic waves reflected in the tunnel and the energy feeding
back from the bridge swinging back and forth during a large earthquake. Series of three-dimensional finite
difference models were developed. Seismic analyses considered both normal and subduction events with a return
period of 250 years. From the results gathered in these analyses, the ground motion modification in the sur­
rounding soil occurs in both the transverse and longitudinal components due to tunnel-bridge interaction. The
tunnel underneath the bridge establishes detrimental soil-structure interaction effects for normal and subduction
events.

1. Introduction interaction in tunnel-bridge systems, including the effects of the distance

between structures and those related to frequency content, duration, and
During strong earthquakes, underground infrastructure such as intensity of the earthquakes. The tunnel-bridge interaction was studied
tunnels modifies the movement of nearby structures. This interaction parametrically throughout the development of a series of
could potentially result in severe damage in densely populated cities, three-dimensional finite difference models within FLAC3D in which the
where tunnels are close to strategic infrastructure like bridges and distance between the bridge and the tunnel was varied. The effect of soil
overpasses, especially if they are on soft clays, such as those found in the nonlinearities was also revised, although these effects were minor for the
Mexico City basin. Therefore, it is vital to consider these effects in their expected shaking level and the high plasticity of Mexico City clay.
design. The ground motion incoherence due to the presence of under­ Neither the tunneling process nor bridge construction were described or
ground structures (i.e., tunnels, cavities, among others) has been studied modeled as this is beyond the scope of this document.
marginally by other researchers [5,14], including some aspects related
to the interface tunnel-soil (i.e., stiffness, cover). However, little 2. Idealized problem
research has been conducted to establish seismic interaction among
on-ground and underground structures [1], especially in densely Tunnel-soil-bridge interaction in soft clays was studied considering
populated areas. Recently [8], conducted a numerical study to investi­ the typology depicted schematically in Fig. 1, using tridimensional finite
gate the ground motion variability associated with the interaction of difference models developed within the program FLAC3D. The tunnel
tunnels and buildings on soft clay deposits, such as those found in width, D, and the urban bridge height, Ho, were set at 11 and 12 m,
Mexico City, considering normal and subduction events. In this research, respectively, to match representative tunnel and bridge typologies in
detrimental effects of the interaction between both structures, as well as Mexico City. For the tunnel-bridge interaction parametrical analyses
with surrounding ones, were established. In a subsequent study [9], presented herein, the distance between the structures varied from 0 to 3
analyzed the effect of the variability of the ground motion using a set of times the tunnel width, D, keeping the tunnel depth constant and equal
records with a broad range of duration, amplitude, and frequency con­ to two times the tunnel width (i.e., 22 m). Fig. 1 compiles the cases
tent. Along this line, the work presented in this technical note builds considered for the tunnel-bridge interaction system. Details of the tunnel
upon those investigations. It continues the study of the seismic and bridge’s characteristics can be found in Ref. [10].

* Corresponding author.
E-mail address: [email protected] (J.M. Mayoral).

https://doi.org/10.1016/j.soildyn.2022.107619
Received 7 July 2022; Received in revised form 11 October 2022; Accepted 21 October 2022
Available online 11 November 2022
0267-7261/© 2022 Elsevier Ltd. All rights reserved.
J.M. Mayoral et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107619

3. Soil profile calibrate a high plasticity clay constitutive model. Therefore, the hys­
teretic model in FLAC3D, denominated as sig3, was used to deal with
The study site is in the so-called Zone IIIb in Mexico City, composed modulus stiffness degradation and damping variation during the seismic
of high plasticity clays. This site corresponds to the reference case event. This model considers an ideal soil, where the stress depends only
analyzed by Ref. [13] and calibrated by Ref. [12]. However, due to the on the deformation and not the number of cycles. With these assump­
lack of detailed geotechnical information, we use this profile to study the tions, an incremental constitutive relationship of the degradation curve
seismic tunnel-bridge interaction on soft clays and continue with the can be described by τn/γ = G/Gmax, where τn is the normalized shear
previous research presented by Refs. [8,9]. stress, γ is the shear strain, and G/Gmax is the normalized secant
modulus. The sig3 model is defined according to Eq. (1):
4. Seismic environment
G a
= ( L− x0
) (1)
Gmax 1 + exp −
Following the research presented by Ref. [9]; the seismic b

tunnel-soil-bridge interaction was analyzed considering the set of six

Where L is the logarithmic strain defined as L = log10(γ), and the pa­
strong ground motions recorded in firm soil or rock with various dura­
rameters a, b, and x0, used by the sig3 model, were obtained by an
tions and frequency contents presented previously by Ref. [9]. As rec­
iterative approach in which the modulus degradation curves were fitted
ommended in the Mexico City building code [11], the seismic
with the model equations. The corresponding damping is given directly
environment was established through uniform hazard spectra, UHS,
by the hysteresis loop during cyclic loading. For the cases studied herein,
developed for return periods of 250 years, considering subduction and
the parameters “a,” “b,” and xo vary from 1 to 1.014, − 0.46 to − 0.55,
normal events, according to the seismogenic activity that controls
and 0.2 to − 1.5, respectively. Non-linear soil behavior is a function of
Mexico City’s seismic risk. To develop acceleration time histories whose
the shaking level, which, if high, leads to shear stiffness degradation and
response spectrum reasonably matches the design response spectrum for
damping increase. The non-linear analyses considered a flexible base at
the return period of analysis (i.e., T = 250 years), these time histories,
the bottom of the model. Fig. 2 shows the numerical model used in the
usually called seed ground motions, were modified using the spectral
tunnel-bridge seismic interaction analysis. Free field boundaries imple­
matching method proposed by Ref. [7] and modified by Ref. [2].
mented in FLAC3D were utilized along the model edges to prevent energy
reflection and represent free field conditions. The model for the tunnel-
5. Numerical model
bridge interaction is composed of 123,138 solid elements with 141,582
nodes. The soil and the bridge foundation box were modeled with solid
Seismic tunnel-bridge interaction analyses were conducted by a se­
elements, the tunnel lining using shell elements, and the bridge super­
ries of three-dimensional finite difference models developed in the
structure and piles were idealized by beam and pile elements, respec­
program FLAC3D [6]. Considering that the high plasticity of Mexico City
tively. As previously used by Ref. [9] for analyses of building-tunnel
clays exhibits a quasi-linear G/Gmax behavior, even for shear strains up
interaction, the same synthetic ground motion was applied to the two
to 0.1%, with a slight damping increase, equivalent linear analyses were
orthogonal horizontal directions to explore a potential extreme event.
deemed appropriate to represent soil non-linearities, particularly in the
Although rigorously this condition is not likely to happen in reality, it is
free field. However, due to the shaking’s level, some degree of soil
standard practice to consider the full magnitude of the accelerations of
non-linearities associated with ground deformations are expected to
the synthetic time history in one direction, and a reduced magnitude of
occur, especially near and at the tunnel-soil and bridge-pile-soil in­
the same synthetic ground motion in the orthogonal component [3,4].
terfaces. Thus, for comparison purposes, fully non-linear site response
Both normal and subduction events were used in the analyses to account
analyses were also carried out to study these effects. Although several
for the variation in frequency content and duration.
constitutive models have been developed to account for non-linearities
to date, there is a lack of enough experimental data to develop and

Fig. 1. Schematic representation of the idealized problem for the tunnel-bridge interaction system.

2
J.M. Mayoral et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107619

Fig. 2. Numerical model for tunnel-bridge seismic interaction system.

6. Tunnel – bridge seismic interaction elastic period of the soil, Tpe = 0.85 s (i.e., 1.17 Hz); 3) and the
maximum nonlinear periods for each seismogenic source in the ampli­
Following the methodology proposed by Ref. [9]; the results are fication function computed for each case for the corresponding equiv­
presented in terms of the “tunnel factor for the bridge,” Tfo, defined as alent linear properties, Tpnl = 1.73 s (i.e., 0.58 Hz) for normal events,
the transfer function computed between the upper deck and foundation and 2.0 s (i.e., 0.5 Hz) for subduction earthquakes. It is also clear that
without the tunnel, TF U–F, divided by the same transfer function with these effects decrease in the frequency range of interest when the tunnel
the tunnel, TF U-FT. These relationships based on the transfer functions moves away from the bridge (Case B), leaving only beneficial interaction
allow the assessment of amplification potential and energy distribution for frequencies greater than about 1.45 Hz. Similar trends were observed
within the frequency content of interest. Values of Tfo greater than one for Cases C and D.
corresponds to beneficial interaction (i.e., the bridge amplification po­ The observed detrimental interaction impacts the maximum relative
tential decreases due to the tunnel presence). In contrast, values smaller displacement of the upper deck (Fig. 5), computed as the maximum
than one indicate detrimental interaction (i.e., the bridge amplification absolute value of the instantaneous relative displacement between the
potential increases due to the tunnel presence). These results are pre­ center of the upper deck and the base of its box foundation for equivalent
sented for all the scenarios considered, for both the transverse (Fig. 3) linear (solid bars) and nonlinear (hatched bars) analyses. In general, the
and longitudinal (Fig. 4) components of movement and equivalent linear maximum relative displacement is more significant in the longitudinal
(solid lines) and nonlinear (dashed lines) analyses. For both types of (y) direction than in the transverse (x) direction, due to the different
constitutive modeling and most of the cases analyzed, there is a clear cross-section of the column. For the subduction earthquakes, the
amplification potential (i.e., detrimental interaction) on the bridge due maximum equivalent-linear relative displacement in the transverse di­
to the tunnel in case A (i.e., tunnel underneath the bridge) for the fre­ rection increases by 28% concerning the case without a tunnel when the
quency range from 1.0 to 3.0 Hz, which include: 1) the fundamental structures coincide (i.e., case A) and decreases between 8 and 10% as the
periods of the structure in the transverse and longitudinal directions, To distance between the tunnel and the bridge increases (i.e., cases C and
= 0.4 and 0.6 s (i.e., 2.5 and 1.6 Hz), respectively; 2) the predominant D). In the nonlinear analyses, the change in the maximum relative

Fig. 3. Relative amplitude of the transfer functions between the upper deck and foundation with and without the tunnel for the transverse direction, using (a)
equivalent linear and (b) non-linear analyses.

3
J.M. Mayoral et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107619

Fig. 4. Relative amplitude of the transfer functions between the upper deck and foundation with and without the tunnel for the longitudinal direction, using (a)
equivalent linear and (b) non-linear analyses.

Fig. 5. Maximum relative displacement computed between the upper deck and the foundation, for the equivalent linear (solid bars) and non-linear (hatched
bars) analyses.

Fig. 6. Shear force, bending moment distribution, and deflection of the most demanded pile during the strongest earthquake, using (a) equivalent linear and (b) non-
linear analyses.

4
J.M. Mayoral et al. Soil Dynamics and Earthquake Engineering 164 (2023) 107619

displacement is minor, increasing up to 25% in case A and decreasing to equivalent linear analyses.
about 5% as separation grows. For normal earthquakes, the maximum
relative displacement in the transverse direction exhibited similar Credit author statement
behavior to subduction events.
The ground motion incoherence due to the tunnel proximity in­ Juan Manuel Mayoral: Conceptualization, Methodology, Re­
creases the rotation of the bridge box foundation and, consequently, the sources, Writing – Original Draft, Project administration, Funding
internal forces in the piles. Fig. 6 depicts the bending moment distri­ acquisition. Daniel De La Rosa: Software, Investigation, Supervision,
bution, and deflection of the most demanded pile during the strongest Writing - Review & Editing, Mauricio Alcaraz: Software, Validation,
earthquake, using equivalent linear and non-linear analyses. The tun­ Visualization, Enrique Barragan: Formal analysis.
nel’s proximity increases the bending moment and shear force on most
piles, compared to the case without a tunnel in both the equivalent
linear analyses and the nonlinear analyses. This tunnel-bridge interac­ Declaration of competing interest
tion effect is seen clearly in the bending moment and is mostly inde­
pendent of frequency content and seismogenic zone. This effect reduces The authors declare that they have no known competing financial
when the tunnel is away from the bridge. The most significant increase interests or personal relationships that could have appeared to influence
in the internal forces of the pile, which occurs in the first third of its the work reported in this paper.
length, is inertial since it is produced by the forces transmitted by the
box foundation as it rotates. Due to the incoherent soil motion, there is Data availability
also a minor kinematic increase in the internal forces acting in all the
pile length. This effect reduces when the distance between the tunnel Data will be made available on request.
and the bridge is larger than two tunnel diameters.

References
7. Conclusions
[1] Abate G, Corsico S, Massimino MR. «Behaviour of coupled tunnel-soil-aboveground
This technical note presents the numerical study of a representative building systems in seismic conditions evaluated by means of parametric analyses,»
tunnel-soil-bridge system often found in the soft clays in Mexico City. A de Proceedings of the 7th international conference on earthquake geotechnical
engineering. Rome, Italy: ICEGE 2019); 2019. 2019.
series of three-dimensional finite-difference models developed in [2] Abrahamson NA. State of the practice of seismic hazard evaluation. Melbourne,
FLAC3D were used to analyze the effect of the underground structure on Australia: s.n.; 2000. p. 659–85.
the seismic performance of a bridge, considering: the tunnel’s proximity [3] ASCE. Minimum design loads for buildings and other structures. ASCE/SEI
Standard 7–10. American Society of Civil Engineers; 2010.
to the bridge; the frequency content of the earthquake; and the influence
[4] Fema P-1050-1/2015. NEHRP recommended seismic provisions for new buildings
of soil nonlinearities. For the specific tunnel-soil-bridge system and other structures.
analyzed, the tunnel leads to ground motion incoherence, changing the [5] Göktepe F. Effect of tunnel depth on the amplification pattern of environmental
nearby soil’s frequency content and spectral acceleration. On the other vibrations considering the seismic interactions between the tunnel and the
surrounding soil: a numerical simulation. Revista de la Construcción 2020;19(2):
hand, although the piled box foundation of the bridge attenuates the free 255–70. 2020.
field ground motion due to its stiffness in contrast with the surrounding [6] Itasca Consulting Group. FLAC3D, fast Lagrangian analysis of continua in 3
soil, both the relative displacement and the rotation of the foundation dimensions, user’s guide. 2009.
[7] Lilhanand Ky Tseng WS. Development and application of realistic earthquake time
increase and, with them, the structural demand on the superstructure. histories compatible with multiple damping resp spec, vol. II. Proc 9th WCEE; 1988.
Therefore, depending on the frequency content of the seismic excitation, p. 819–24.
it is expected to have beneficial interaction in certain subduction [8] Mayoral JM, Mosqueda G. Seismic interaction of tunnel-building systems on soft
clay, vol. 139. Volumen: Soil Dynamics and Earthquake Engineering; 2020,
earthquakes and detrimental in normal events. An effect of utmost 106419.
importance is the increment in bending moment and shear force acting [9] Mayoral JM, Mosqueda G. Foundation enhancement for reducing tunnel-building
on the piles in the foundation of the bridge, which can reach up to 25% seismic interaction on soft clay. Tunn and Und Space Tech 2021;115:104016.
[10] Mayoral JM, De La Rosa D, Alcaraz M. Seismic performance of tunnel-building-
for the cases analyzed. These effects are more significant when the urban bridge systems in soft clays. Submitted to Tunn and Und Space Tech 2022.
tunnel is underneath the bridge, nevertheless, decrease as the tunnel [11] NTCS. Normas Técnicas Complementarias para Diseño por Sismo. Gaceta Oficial de
gets away from the bridge and increase when the excitation frequency la Ciudad de México; 2020.
[12] Romo MP. Clay behavior, ground response and soil-structure interaction studies in
content matches the site’s fundamental and bridge period. This increase
Mexico city. Int Conf on Recent Advances in Geotechnical Earthquake Eng and Soil
will be significant when new underground structures cross under old Dyn 1995.
structures supported by friction piles. Finally, for the specific case [13] Seed HB. Relationships between soil conditions and earthquake ground motions in
analyzed, the difference between equivalent linear and nonlinear ana­ Mexico city in the earthq of sept 19, 1985. Berkeley, California: College of Eng, UC;
1987.
lyses was much more quantitative than qualitative, following similar [14] Tsinidis G, Pitilakis K, Trikalioti AD. Nume sim of round robin num test on tunnels
trends. As expected, ground shaking is lower in nonlinear than using a simplified kin hard model. Acta Geotechnica 2014;9:641–59. 2014.