Computers and Geotechnics 120 (2020) 103385

Contents lists available at ScienceDirect

Computers and Geotechnics

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

Research Paper

Analytical study on interaction between existing and new tunnels parallel T

excavated in semi-infinite viscoelastic ground
H.N. Wanga,b, , X. Gaoa, L. Wua, M.J. Jiangb,c
⁎

a
School of Aerospace Engineering and Applied Mechanics, Tongji University, Shanghai 200092, China
b
State Key Laboratory of Disaster Reduction in Civil Engineering, Tongji University, Shanghai 200092, China
c
Department of Civil Engineering, Tianjin University, Tianjin 300072, China

ARTICLE INFO ABSTRACT

Keywords: A new tunnel is usually excavated in close proximity to existing ones, which leads to significant negative in-
Viscoelasticity fluence on the existing tunnels. Moreover, the rheology of the ground probably induces quite different time-
Shallow tunnels dependent ground deformation if the new tunnel is excavated at different time. This study focuses on the in-
Analytical solution teraction between the existing and the new tunnels which excavated in rheological ground at a shallow depth.
Existing and new tunnels interaction
Through the strict derivation, a new analytical solution is proposed for ground stress and displacement induced
by the interaction of new and existing tunnels. The ground rheology, excavation delay of the new tunnel, tunnel
size and various tunnel arrangement, are all taken into account.
The complex variable theory combined with the extension of corresponding principle are employed in the
derivation. By deriving the potentials in complex variable theory for the problems in all the excavation stages,
the time-dependent stresses and displacements are finally addressed for the whole excavation process, where the
ground is simulated by any linear viscoelastic models. To verify and validate the analytical solutions, the ana-
lytical solution is compared with numerical results under simplified and complex ground conditions, which
shows good consistency except the solution for Case 1 (considering gravity gradient). A parametric study is
finally preformed to find the influence of excavation time and location of the new tunnel, the tunnel spacing and
relative size of the new tunnel, on stresses/displacements around tunnels and surface settlements. The results
show that the excavation time of the new tunnel (t2) significantly influence the additional displacements around
the existing tunnel which is a decrease exponential function of t2; when the distance from center to center is
larger than 2.5R1 (2.0R1), the interaction between two tunnels can be neglected from perspective of displace-
ment (stress).

1. Introduction ground responses when the new tunnel is excavated at different time. In
urban areas, tunnels are generally constructed at a shallow depth, and
New tunnel excavation in close proximity to the existing ones, as greatly influenced due to the presence of ground surface. and This study
well as the sequential excavation of twin tunnels, are commonly en- focuses on the time-dependent ground responses induced by the inter-
countered in underground constructions. The new tunnel excavation action between existing and new tunnels at shallow depth.
induces additional displacement and stress around existing tunnels, Numerical methods are usually employed in detailed design to
which probably results in instability of the existing tunnels [1–3]. predict the ground responses induced by a new tunnel excavation near
Furthermore, the stress and displacement around the new tunnels are the existing ones, with consideration on complex geological conditions
also quite different from those in single tunnel problem, due to the and sequential excavation [6–10]. Numerical method provides very
influence of the existing tunnel boundaries. Therefore, the interaction helpful results but requires longer running-times, especially when
between existing and new tunnels should be exactly taken into account complete parametric analyses are performed. Empirical formulas with
in assessment of tunnel stability. simple expressions are also widely used in engineering practice
Rock or soil probably exhibits time-dependent behaviors (rheology), [11–13]. For example, superposition technique on the Gaussian equa-
which induces gradual deformation over time even after the completion tion suggested by Peck [11] is one of the most popular empirical for-
of the tunnel excavation [4,5]. Ground rheology induces quite different mulas based upon field observations to predict the soft ground surface

⁎
Corresponding author at: School of Aerospace Engineering and Applied Mechanics, Tongji University, Shanghai 200092, China.
E-mail address: [email protected] (H.N. Wang).

https://doi.org/10.1016/j.compgeo.2019.103385
Received 29 August 2019; Received in revised form 5 November 2019; Accepted 8 December 2019
Available online 24 December 2019
0266-352X/ © 2019 Elsevier Ltd. All rights reserved.
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

movement induced by shallow twin tunneling [12,13]. However, the

empirical equations fail to consider the time effect induced by the
rheology properties of ground and the sequential excavation of two
tunnels. Also the empirical formulas can only predict the surface set-
tlement instead of the displacement and stress fields of the whole
ground.
In contrast, the simplified analytical models can be employed to
efficiently obtain initial estimation of the design parameters in pre-
liminary design. The analytical models provide the stress and dis-
placement field of the whole ground through strict mathematical de-
rivation, adopting all of the concerned parameters. They also provide
an approach to gain deep insight into the mechanical mechanisms of Fig. 1. Geometry of the problem, and reference coordinate systems.
engineering problems.
Because of the difficulty in analytical derivation, the ground is as-
medium/stiff clay, is in a state of small deformation. The ground
sumed to be elastic in most of analytical research for two-tunnel (hole)
rheology is accounted for by various linear viscoelastic models [28].
problem in infinite or semi-infinite ground. Aiming at the stress around
Assumed that the cross-section excavation is instantaneous. To
two holes in infinite elastic plane, the solutions were proposed by in-
tackle the problem as two dimensional plane-strain one, two advance-
troducing the Airy stress functions in bipolar coordinates in many stu-
ment parameters 1 (t ) and 2 (t ) (0 < 1 ( 2 ) 1), are introduced to ac-
dies [14–18]. Based on the bipolar coordinates and Schwartz alter-
count for the progressive release of the stresses cause by the existing
nating method, the elastic analytical solutions were presented by Kooi
and second tunnel longitudinal advancement, respectively. i (t ) , i = 1
and Verruijt [16] for displacement and stress caused by the interaction
and 2, is a function of the variation of the radial displacement with the
of twin tunnels in deep depths. Hoang and Abousleiman [17] presented
distance z between the tunnel face and the cross-section considered,
the stress solution for an infinite plate containing two equal/unequal
that can be determined by in-situ measurements, or calculated nu-
circular holes, subjected to general stresses at infinity and internal
merically [29–31]. i (t ) equal to 1.0 in case of the distance z being
pressure along hole boundaries. For the holes in half-space, Spencer and
sufficient large.
Sinclair [19] employed a sequence of Airy stress function to derive an
According to the aforementioned assumptions, the equivalent plane-
analytical solution for twin circular holes under gravitational load.
strain problem in the plane of tunnel cross-section of the tunnels can be
These approaches are difficult in providing the displacement fields.
simplified, as illustrated in Fig. 1. Both the Cartesian coordinates
The complex variable theory [20–25] is powerful in solving the
(x1, y1) , (x2 , y2 ) and polar coordinates (r1, 1) , (r2, 2) are employed in the
elastic cases of non-circular openings or the cases with multiply con-
following derivation of analytical solutions. A tension-positive notation
nected regions. Based on complex variable theory, Manh et al. [20]
is used throughout this study.
presented the solutions for the stresses and displacements around two
According to the start excavation time of two tunnels, the excava-
deep circular tunnels in an elastic ground. For non-circular holes, based
tion process is divided into two stages in the derivation presented
on the Schwarz alternating method, the stress concentration factors
herein: in the first tunnel excavation stage spanning from t = t1 to t = t 2 ,
were obtained by Ukadgaonker and Patil [22] for two unequal elliptical
only the first tunnel is excavated, as shown in Fig. 2(a). In the second
holes with different sizes, subjected to internal pressure and shear
tunnel excavation stage spanning from t = t 2 onwards, both the two
stresses along hole boundaries. Zhang et al. [23,24] provided the ac-
tunnels are excavated, as shown in Fig. 2(c).
curate stress solution for two/multiple elliptical holes in infinite region
Note that t2 may be either much larger than or equal to t1 so that the
by using Schwarz alternating method and complex variable method. In
sequential excavation process is accounted for including the excavation
recent analytical research [26], the stresses and displacements were
of both tunnels occurring at the same time. In the first tunnel excava-
provided around two closely located circular tunnels in deep ground,
tion stage (Fig. 2a), x(2 1) and y(2 1) are the tractions acting on the
considering the rock viscoelasticity and the construction sequence.
future boundary of the second tunnel; whereas no tractions are exerted
However, the solution was only applied to deep tunnel analysis. For
on the internal tunnel boundaries in the second tunnel excavation stage
twin tunnels in an elastic half space, complex variable method com-
(Fig. 2c). Therefore, according to the superposition principle (viscoe-
bined with Schwarz alternating method are usually employed: Fu et al.
lasticity), the incremental mechanical responses after the second tunnel
[25] presented the solutions for stress and displacement around shallow
excavation can be addressed by the model in Fig. 2(b).
buried twin parallel tunnels, taking into account the prescribed uniform
radial displacement along tunnel boundaries; Wang et al. [27] proposed
the elastic solutions for shallow twin tunnels subjected to arbitrary 3. Derivation of analytical solutions
distributed surcharge loads.
In summary, the aforementioned analytical studies fail to consider 3.1. Formulation for the general viscoelastic problem
the ground rheology and the sequential excavation of tunnels, or the
tunnels were set in infinite viscoelastic ground. In this study, the ana- The different stress-strain behavior of linear viscoelastic model can
lytical solution for ground displacement and stress after new tunnel be schematized by a number of springs and dashpots connected either
excavation will be strictly derived, considering the ground viscoelasti- in series or in parallel. The methodology is provided in the literature [5]
city, the delay excavation of new tunnels and tunnels interaction. We for solving a general viscoelastic problem involving time-dependent
expect that the solutions can help reveal the mechanical mechanism boundaries.
and the law of the interaction between the new and existing tunnels. In a sequential excavation, the state of stress is dependent of the
excavation steps carried out over time, and the displacements in vis-
2. Problem definition coelastic medium depend on the entire previous stress history. For a
general problem, l loads are assumed to exert on the structure at dif-
Prior to tunnel excavation, the semi-infinite ground without any ferent time before the generic time t , i.e., the k–th load (k = 1,2,…., l)
tunnels is subjected to gravity induced initial stresses. Two parallel is applied on the structure at time tbk and removed at tmk . According to
circular tunnels, i.e., the first (existing) tunnel and the second (new) one the superposition principle for viscoelastic case along with complex
are then successively begin excavated at time t = t1 = 0 and t = t2 , re- variable theory [32], the total displacement at the generic time t can be
spectively. In this study, the ground which composes of rock or calculated as follows:

2
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

Fig. 2. Schematic representation of the stress boundary conditions for calculation of (a) stresses and displacements before excavation of second tunnel; (b) stresses
and displacements induced by excavation of second tunnel; (c) total stresses and displacements. The superposition relation is also shown.

uxv (z , t ) + iu yv (z , t ) u1vx (z , t ) + i u1vy (z , t )

l
1 1 ( ¯1)
Tk
(k ) 1 t 1 (z ) ¯
= I (t ) 1 (z , ) d = ( 1)
(z ) 1( ) I (t )d z + ( 1)
(z )
2 k=1
tbk 2 2 0 2 z
l (k )¯
1 (z , )
Tk t
(k ) ¯(z, ) d 1( ) H (t )d
H (t ) z + 1 0 (4)
tbk z
k=1 (1)
According to Eq. (3) and considering the initial stress fields, the
where Tk = min{tmk, t } . 1(k ) and 1(k ) are the potentials corresponding to total stresses can be expressed as follows:
the case subjected to k–th load in the loading period [tbk, tmk ]; z = x + iy
and i = 1 ; e (z¯, t ) are the conjugates of the complex function v
1x (z , t) p0x ( 1) (z ) 2 ( 1) (z ) ( 1) (z )

e = e (z, t ) , and v = p + Re 2 1 (t ) 1 (t ) z +
z2
1y ( z , t) 0y z z

1 1 v
2 ( 1) (z ) ( 1) (z )
H (t ) = L , 1xy (z , t ) = Im z 1 (t )
z2
+ 1 (t ) z
sL [G (t )]

1 1 3L [K (t )] + 7L [G (t )] (5)
I (t ) = L (plane strain problem),
sL [G (t )] 3L [K (t )] + L [G (t )] where p0x and p0x are horizontal and vertical initial stresses induced by
(2) gravity, respectively.
where L [f (t )]) with respect to variable s is defined in the Laplace
transform of the function f (t ) , andL 1 [g (s )] indicates the inverse La- 3.3. Solutions for the second tunnel excavation stage
place transformation of g(s); G(t) and K(t) represent the shear and bulk
relaxation moduli of viscoelastic model, respectively. The expressions In order to obtain the time-dependent ground responses, the var-
of H(t) and I(t) for generalized Kelvin and Poynting-Thomson viscoe- iation of boundary conditions during the whole excavation stage should
lastic models can be found in Table 3 [33]. be clarified. A model of two holes in half-plane is assumed with their
The stresses are addressed by exploiting the principle of super- boundaries loaded by various tractions during different excavation
position as: stages: (a) during first excavation stage, the ground is subjected to
v gravity, and the first tunnel boundary is stress-free. The future
x (z , t) l (k ) (z , t ) 2 (k ) (z , t ) (k ) (z , t )
v = Re 2 z¯ + , boundary of the second tunnel is subject to surface forces x(2 1) and
z2
y (z , t) z z
k=1 (2 1)
y (see Fig. 2a) determined by the stress field in Eq. (5); (b) during
l 2 (k ) (z , t ) (k ) (z , t ) second tunnel excavation stage, the ground is subject to gravity and the
v
xy (z , t ) = Im z¯
z2
+ z
. two hole boundaries are both stress-free (see Fig. 2c). If considering the
k=1 (3)
progressive stress release, this can be equivalent to exerting the trac-
(k )
(z , t ) t [tbk , tmk ] tions, (2 1)
2 (t ) x and (2 1)
2 (t ) y on second tunnel boundary (see
where (k )
(z , t ) = 1
and (k )
Fig. 2b) from t2 to generic time t on the bases of the boundary conditions
0 t [tbk, tmk ]
in Fig. 2(a).
(k )
(z , t ) t [tbk , tmk ] According to Eq. (1), the additional displacements at time t
(z , t ) = 1
; Re[·] and Im[·] denote the real and
0 t [tbk , tmk] (t [t2, ) ) occurred after the second tunnel excavation can be ex-
imaginary component of a generic complex variable [·]. pressed as follows:

3.2. Analytical solutions for the first tunnel excavation stage u2vx (z , t ) + i u2vy (z, t )
= u2(1) v (1) v (2) v (2) v
x (z , t ) + i u 2y (z , t ) + [u 2x (z , t ) + iu 2y (z , t )] (6)
The additional stresses and displacements caused by excavation are
crucial in engineering application. Prior to tunnel excavation, the where the displacements are divided into two parts, i.e., u.(1) v and
gravity-induced initial stresses acted on the future boundaries of the u.(2) v , whose expressions are the follows:
tunnels, e.g., x(1 0) and y(1 0) on the boundary of the first tunnel. The
1 t
additional ground responses induced by the first tunnel excavation can Part 1: u2(1) v (1) v
x + i u 2y = 2 [ 0 1
( ) I (t )d
be determined by the model of the first tunnel subjected to tractions t2 ( 1)
1 ( ) I (t2 )d ] (z )
(1 0)
1 (t ) x and (1 0)
1 (t ) y along tunnel boundary. Based on the 0
1 t
previous research [27], the corresponding elastic complex potentials of [
2 0 1
( ) H (t )d
this problem in case of 1 = 1, ( 1) ( ( 1) ), can be addressed (detailed in ( ¯1) (z )
t2 ¯
Appendix A). According to Eq. (1), the additional displacements at
( 1)
0 1( ) H (t2 )d ] z + (z )
z
generic time t (t [t1 = 0, t2]) of viscoelastic problem can be expressed
as follows: (7)

3
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

1 t
Part 2: u2(2) v (2) v
x (z , t ) + iu 2y (z , t ) = 2
( 2)
(z ) t2 1( ) 2 ( ) I (t )d
( ¯2) (z )
1 t ¯
( 2)
2 t2 1( ) 2 ( ) H (t )d z + (z )
z

(8)
( 2)
and in part 2 denote the potentials for the twin tunnel
( 2)

model in Fig. 2(b) in cases of 1 = 2 = 1.0 . Determination of these

potentials is detailed in Appendix B, where Schwartz alternating
method is introduced to translate the two tunnel problem into a series
of single tunnel problems. Noted that the part 2 displacements are in-
duced by excavation of the second tunnel, and the part 1 displacements
are the additional rheology displacements in this stage induced by the
first tunnel excavation. Therefore, the displacements occurred after
excavation of the second tunnel are significantly dependent of the
whole sequential excavation process of the two tunnels. However for
the elastic problem, the part 1 displacements in Eq. (7) are zero.
The additional stresses occurred during this stage are:
v
2x (z , t) ( 2) (z ) 2 ( 2) (z ) ( 2) (z )
v = Re 2 1 (t ) 2 (t ) 1 (t ) 2 ( t ) z +
z2
2y (z , t) z z

2 ( 2) (z ) ( 2) (z )
v
2xy (z , t ) = Im z 1 (t ) 2 (t )
z2
+ 1 (t ) 2 (t ) z

(9)
Because the effects of longitudinal advancement are investigated
only in case of single tunnels [30,31], the expressions for the ad-
vancement parameters, 1 and 2 , are not available in the literature.
Therefore, in the following of the paper 1 = 2 = 1.0 is assumed. This
means that we consider only cross-sections located at a distance from
both tunnel faces such that three dimensional effects are not felt.

4. Verification and validation of analytical solutions

Fig. 3. Geometry of the domain together with the boundary conditions in nu-
4.1. Verification of analytical solutions
merical simulation for verification of derivation of (a) Case 1 and (b) Case 2.

To verify the analytical solution, the derived solutions are compared

with the results from numerical simulations carried out using the In the numerical simulation, the gravity induced the initial stresses
software ANSYS (Version 14.0, employing the “structure mechanics” are first generated in the half-plane without any holes. Afterwards, the
module). To maintain the consistency with the analytical derivation, all elements inside the left hole are removed to simulate first tunnel ex-
the FEM analyses are performed under plane-stain conditions and the cavation at t = 0, where the stiffness matrix of these elements was
same assumptions in the analytical derivation. multiplied by a coefficient of 10 6 (to deactivate elements for the death
A semi-infinite medium with two circular holes are considered. In capability in ANSYS). Consequently, the elements inside the right hole
the simulation, the generalized Kelvin model with shear modulus are removed after 30 days (t2 = 30 day) to simulate the excavation of
G M = 20MPa (for Maxwell part), G K = 10MPa (for Kelvin part), and second tunnel. The excavation induced incremental displacements and
coefficient of viscosity K = 100MPa· day , is assumed for the ground, stresses can be obtained by subtracting the displacements and stresses
and the Poisson’s ratio v is 0.25. Meanwhile, the unit weight and that occurred before excavation from the total ones. The numerical
lateral pressure coefficient K 0 are 18.5kN m3 and 0.27, respectively. The model has been verified in our previous studies [26,35,36].
effect of longitudinal advancement is not considered, i.e., 1 = 1 = 1.0 . Figs. 5 and 6 show the comparison on additional displacements after
According to the conditions in practical engineering, the comparison for the first tunnel excavation, as well as the total stresses versus time at
the following two cases is performed: points P1- P4 (point locations are shown in Fig. 4), predicted by the
analytical solutions and FEM simulations. Noted that the results from
Case 1:. the variation of vertical gravity loading across the height of analytical and numerical models exhibit a close agreement for both
excavation is considered in the initial stress state; Cases 1 and 2. Thus, the analytical and numerical results are mutually
Case 2:. the variation of vertical gravity loading across the height of verified.
excavation is neglected. The initial stresses along anticipated tunnel
boundary are assumed approximately equal to the stresses at the center
4.2. Validation of analytical solutions
of the tunnel [34].
The geometry and boundary conditions applied in FEM simulation In order to further validate the applicability of proposed analytical
for two cases are shown in Fig. 3: the vertical displacements along the model, the comparisons on surface settlement and displacements along
bottom boundary and the horizontal displacements along left and right tunnel boundaries were performed between analytical solutions and
boundaries are restrained; the two tunnels have the same size and results from finite differential method (FDM), where the more complex
buried depth in verification of Case 1, whereas in Case 2 verification, constitutive relations and geological condition of the ground, are
the size and buried depth of two tunnels are quite different. Fig. 4 adopted in the numerical model.
presents the mesh in the vicinity of the tunnels for two cases, where the Numerical analyses were performed by software FLAC3D for surface
mesh in the yellow area represents the excavated rock or soil. settlement in the Istanbul metro [37]. The numerical results were

4
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

balance out the resultant forces Y (see Eq. (A2) in Appendix A) induced
by the variation of vertical gravity loading (Y is equal to the weight of
excavated rock/soil, the direction is vertical upward), which makes the
Case 2 model being more close to the actual situation than Case 1
model. Therefore, the analytical solutions of Case 2 will be employed in
the following parametric investigation.
By using software FLAC3D, the deformation along the first and the
second tunnel boundary induced by the driving of new tunnel in the
shallow depth was presented in the reference [8]. The geometry of the
problem is shown in Fig. 9 and the parameters of mechanical properties
are presented in Table 2. The comparisons are plotted in Fig. 10 on
radial displacements induced by the second tunnel excavation between
analytical solution of Case 2 and numerical results for various tunnel
spacing. It is noted that the variation pattern of induced radial dis-
placements for all the cases are very similar between analytical and
numerical results, and the magnitude of analytical results is somewhat
larger than numerical one due to the absent of lining in the analytical
model.
From the good agreement between analytical and numerical results
for real engineering, we believe that the proposed analytical solution of
Case 2 can provide reliable prediction of displacement field qualita-
tively.

5. Mechanical mechanism of time-dependency of displacement

Different viscoelastic models can simulate different rheological

characteristics of geo-material [38]. Generalized Kelvin and Poynting-
Thomson viscoelastic models have limited viscous deformation under
constant stress, which are usually employed for simulating the rock
with good mechanical properties or subjected to low stress.
Different from the elastic problems where the additional displace-
Fig. 4. FEM mesh of the vicinity domain: (a) and (b) the mesh for Cases 1 and 2,
ments after the second tunnel excavation are completely comes from
respectively.
the excavation of the second tunnel, the additional displacements are
both influenced by the excavation of the first and second tunnels when
closely consistent with field measurements [37], which proved the the rheology of the ground is considered (part 1 and part 2 displace-
applicability of the numerical model. Fig. 7 shows the geological geo- ments shown in Eqs. (7) and (8). According to the expressions of
metrical conditions. The parallel twin tunnels with diameter of functions H(t) and I(t) provided in Table 3, the results of the four in-
D = 6.16 m, both located at depth of 9.74 m. In the numerical simu- tegrations about t in expressions of part 1 and part 2 displacements are
lation, the elasto-plastic analysis based on Mohr-Coulomb failure cri- presented in Table 4 for the two viscoelastic models ( 1 = 2 = 1.0 ).
terion for the ground was carried out with the shotcrete lining and soil The functional forms of the integration expressions for the two
nails modeled by different elements. The thickness and mechanical models are similar. According to the expressions of J1 and J2, the in-
properties of soil layers are provided in Table 1 [37]. The equivalent fluence of the first tunnel excavation (part 1) on the additional dis-
material parameters employed in analytical solution are determined as placements exponential decreases against time, and is a constant C
follows: when t tends to infinity. The constant C is an exponential function of t2,
Q1 h1 + Q2 h2 + Q3 h3 + Q4 h4 i.e., exp( Ct2 ) , which shows the influence of excavation time of the
Qe = second tunnel on additional displacement is a decrease function. The
h1 + h2 + h3 + h 4 (10)
displacements induced by the second tunnel excavation (part 2) are the
where Qe represents the equivalent parameter (i.e. Young’s modulus, exponential increase functions of (t2-t). Noted the excavation time of
Poisson’s ratio and unit weight) in analytical solution; Qj and the second tunnel has no influence on the final part 2 displacement.
hj (j = 1, 2, 3, 4) represent the material parameters and thickness of soil
layer j in Table 1, respectively. The effect of support, the construction 6. Parametric investigation
process and ground plasticity are neglected in the analytical calcula-
tion. For geo-materials with good mechanical properties and low stresses
Based on these parameters, the final ground settlements obtained by in application, limited viscosity is present thus the Kelvin, generalized
the proposed analytical solutions (for Cases 1 and 2), as well as the Kelvin or Poynting-Thomson viscoelastic models are commonly em-
corresponding FDM results, are plotted in Fig. 8 for various tunnel ployed. In this section, a parametric analysis is carried out considering
spacing. The variation patterns of surface settlement predicted by the effects of the various parameters on the displacements and stresses
analytical model of Case 2 (without gravity variation) are closely con- around the tunnels. The ground rheology is described by the general-
sistent with the FDM results in tendency except for the larger values. ized Kelvin viscoelastic model with constitutive parameters GM GK = 2 .
This may be due to the simplification in the analytical model, e.g., the The displacement in this section is the incremental one occurred after
neglecting of lining support and ground plasticity which were con- the second tunnel excavation, and the stress is the total one.
sidered in the numerical model. 1 = 2 = 1.0 in the following analysis (fast advancement).
In contrast, the variation pattern and magnitude of settlements For the sake of generality, the displacements and stresses are nor-
obtained by analytical model of Case 1 (with gravity variation) are both malized by ur and p0 , respectively, where p0 is the initial vertical stress
very different from FDM results. The reason probably is that: in the real at the depth of first tunnel center; ur is the radial displacement on a
engineering, the weight of liner and shotcrete-bolt support can partly tunnel boundary for a circular tunnel excavated in an infinite elastic

5
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

Fig. 5. Comparison on displacements between analytical and FEM results: (a) and (b) displacements for Case 1; (c) and (d) displacements for Case 2. P1 P4 are points
as shown in Fig. 5.

medium, subjected to hydrostatic far field stress p0 : the first tunnel excavation, called non-simultaneous excavation;
(c) t2 = 5.0TK , i.e., the second tunnel is excavated after a very long
p R1
ur = 0 time of the first tunnel excavation, called stable state excavation.
2Gs (14)
Meanwhile, the buried depths and radii of tunnels are H1 = H2 = 2R1
where R1 is radius of the first tunnel; Gs = GM GK (GM + GK ) is the and R1 = R2 , respectively; the spacing between twin tunnels is c = R1. In
permanent shear modulus. Other parameters indicated dimension are the following analysis, the additional displacements occurred after ex-
normalized by R1. The generic time t in viscoelastic cases is normalized cavation of the second tunnel, and the total stress of the ground, are
by the retardation time TK (TK = K GK for the generalized Kelvin provided by analytical model.
model). The sign convention adopted in this section is the same as that The additional displacements at the given points versus time are
in Section 4. plotted in Fig. 11 for stable state excavation case. The plots illustrate
that the displacements increase with time and tend to be stable ap-
proximate at time t TK = 8 . Because the final stable displacements are
6.1. Influence of the excavation time of the second tunnel
the most significant in tunnel design, these displacements are adopted
in the following parametric analysis.
In this sub-section, three cases with following excavation time of
In Fig. 12 the final surface settlements occurred after the excavation
second tunnel are considered:
of the second tunnel against x1 are presented for the aforementioned
(a) t2 = 0 , i.e., the second tunnel is excavated instantaneously after
three cases. The figure shows that the settlements along the ground
the first tunnel excavation (simultaneous excavation), thus only elastic
surface at the region over the first tunnel are quite different when the
deformation has been occurred before the second tunnel excavation;
second tunnel excavated at different times, e.g., they obviously decrease
(b) t2 = TK , i.e., the second tunnel is excavated after a certain time of

6
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

Fig. 6. Comparison on stresses between analytical and FEM results: (a) and (b) stresses for Case 1 and Case 2, respectively.

Fig. 7. Geological properties and geometry of twin tunnel problem in validation of analytical model: (a) geological profile of longitudinal cross section; (b) geometry
of twin tunnels.

Table 1
Thickness and properties of soil layers in Zone C of project area [37].
Layer Soil type Thickness (m) Unit weight (kN/m3) Mohr-Coulomb shear strength parameters Young’s modulus (MPa) Poisson’s ratio

Cohesion (kPa) Friction angle (deg)

Layer 1 Artificial filling 2.95 18 1 10 5 0.4

Layer 2 Sand 10.6 17 1 25 15 0.35
Layer 3 Suleymaniye formation 11.32 18.9 20 14 38 0.33
Layer 4 Trakya formation Base 25 80 25 60 0.2

7
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

Fig. 8. Comparison on surface settlement versus x1 between analytical solution for Case 1/Case 2, and FDM results [37]: (a) and (b) the settlements in case of tunnel
spacing c = 5D and c = D , respectively.

that in Section 5, and the function form of displacement variation with

time can be clearly found through the analytical solutions.
The hoop stresses along first tunnel boundary versus 1 are plotted
in Fig. 14 for twin tunnels and single tunnel problem. The interaction of
the two tunnels causes the stresses around 1 = 0° in twin tunnels pro-
blem larger than that in single tunnel problem, and the maximum dif-
ference is up to 20% of the maximum stresses in single tunnel problem.
Meanwhile, from Fig. 13(b) it is seen that the maximum difference of
final displacements between twin tunnels and single tunnel problem is
occurred around 2 = 135° and the difference is up to 22% of that in
single tunnel problem.
Fig. 9. Geometry of the tunnels in FDM simulation [8].

Table 2 6.2. Influence of location of the second tunnel

Soil properties in the project area [8].
The new tunnel can be constructed at different position near the
Unit Mohr-Coulomb shear Young’s Poisson’s ratio
weight strength parameters modulus
existing tunnel in practical engineering, which induces the quite dif-
(kN/m3) (MPa) ferent ground responses. According to the relative position of the two
Cohesion Friction tunnels, the arrangement angle is defined in Fig. 1. In this sub-section,
(kPa) angle (deg) the first tunnel is fixed at a depth H1 = 2R1, and the following five cases
Soil 18.5 0 37 150 0.3
with different location of the second tunnel are considered: (1)
= 15° ; (2) = 0° ; (3) = 30° ; (4) and (5) = 60° = 90° . It is as-
sumed that the size of the two tunnels are equal. The tunnel spacing is
c = R1 and the second tunnel is excavated after a very long time of the
with the delay time of second tunnel excavation, whereas the settle- first tunnel excavation.
ments are almost the same in the three cases at the region over the The final additional displacements along tunnel boundaries versus
second tunnel.
1 or 2 are plotted in Fig. 15 for the five cases, where the variation and
Fig. 13 presents the final additional displacements along tunnel magnitude of displacements observed are both quite different, espe-
boundaries versus 1 or 2 for various excavation times of the second cially for those along the first tunnel boundary. Fig. 15(a) and (b) in-
tunnel. The differences of displacement among these cases are obvious dicate that the displacements along the first tunnel boundary are a great
along the first tunnel boundary, however little difference is observed of fluctuation, and are notable larger in the cases with = 60° and 90° . The
displacements along the second tunnel boundary. Therefore, it is con- radial displacements show a peak value at 1 = 330° 300° 270°when the
cluded that the excavation time of the second tunnel only significantly arrangement angle is 30° 60° 90° , respectively. It is emerged from
influences the additional displacements around the first tunnel. The Fig. 15(c) and (d) that the variations of displacements along the
earlier the second tunnel excavated, the larger the displacements boundary of second tunnel are similar in the five cases, except for the
around the first tunnel are. The final displacements around the second magnitudes. The maximum radial displacement along second tunnel
tunnel are almost unaffected by the excavation time of the second boundary occur at 2 = 90° in the case with = 15° , due to the closer
tunnel. It can be noted that the above conclusions are consistent with distance from second tunnel center to ground surface. The maximum

8
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

Fig. 10. Comparison on deformations of tunnel boundaries between analytical and FDM results [8]: (a) and (b) radial displacements along the first tunnel boundary
obtained by analytical and numerical results, respectively; (c) and (d) radial displacements along the second tunnel boundary obtained by analytical and numerical
results, respectively.

9
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

Table 3
Expressions of H(t) and I(t) for the two viscoelastic models (modified from [33]).
H(t) I(t)

Generalized Kelvin model

( )
1 1 GK 6GM 2
(t ) + exp t 6
GM H (t ) + (t ) + 2·
K K 3K e + GM K (3K e + GM )
3K e GK + GM (3K e + GK )
exp t
K (3K e + GM )
Poynting-Thomson model
( ) exp ( ) (t) +
2 6GP2
1 GP GP GH 6 1
t + 2·
P GP + GH (GP + GH) P 3K e + GP + GH GP + GH P (3K e + GP + GH)

( ) exp
1 2
+ (t ) exp
GP (3K e + GH )
t +
1 GP GP GH
t
GP + GH
P (3K e + GP + GH) P GP + GH (GP + GH) P

Table 4
Integration results for the expressions of displacements for Generalized Kelvin and Poynting-Thomson viscoelastic models.
Displacement occurred after the second tunnel excavation = Part 1 + Part 2 displacements

Part 1 displacements, induced by the first tunnel excavation

( ¯1) (z )
u 2(1) v (1) v 1 ( 1) 1 ¯ (z ) , with J (t ) =
+ ( 1)
t t2
) d ; J2 (t ) =
t t2
)d .
x + i u 2y = J1 (t )
2
(z ) J2 (t ) z
2 z
1 0
I (t )d 0
I (t2 0
H (t )d 0
H (t2

Part 2 displacements: induced by the second tunnel excavation

( ¯2) (z )
1 1 ¯ t t
u2(2) v (2) v
x (z , t ) + iu 2y (z , t ) = J3 (t )
( 2)
(z ) J4 (t ) z + ( 2) (z ) , with J3 (t ) = t2
I (t ) d , J4 (t ) = t2
H (t )d .
2 2 z

Integration Generalized Kelvin model Poynting-Thomson model

J1 (t ) J2 (t )
BK
exp( CK t )[1 exp( CK (t2 t ))] BP2
CK J2 (t ) 1 exp( CP2 t )[1 exp( CP2 (t2 t ))]
CP

J2 (t ) 1
BP
1 GK GK
exp t 1 exp (t2 t) 1 exp( CP1 t )[1 exp( CP1 (t2 t )]
GK K K CP

J3 (t ) 1
BP BP2
1 GK 1 BK
1 exp t) + + [1 exp( CK (t t2))] + AK 1 [1 exp( CP1 t )] + AP1 + 1 [1 exp( CP2 (t t2))] + AP2
GK K GM CK CP CP

J4 (t ) 1
BP
1 GK 1
1 exp (t t2) + 1 [1 exp( CP1 (t t2))] + AP1
GK K GM CP
2
6GM
6 3K e GK + GM (3K e + GK )
Generalized Kelvin: AK = , BK = 2,
CK = ;
3K e + GM K (3K e + GM ) K (3K e + GM )

( ) ,B 6GP2
2
1 6 1 GP GP GH GP (3K e + GH )
Poynting-Thomson: AP1 = , AP2 = , BP1 = 2
P = 2
, CP1 = , CP2 = .
GP + GH 3K e + GP + GH P GP + GH P (3K e + GP + GH) (GP + GH) P P (3K e + GP + GH)

Fig. 11. Normalized displacements versus time at points with 1 = 0° 90° , 2 = 180° along tunnels boundaries and O1 for stable state excavation: (a) and (b) the
horizontal and vertical displacements, respectively.

10
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

Fig. 14. Normalized hoop stresses along the first tunnel boundary versus 1 for
Fig. 12. Surface settlements versus x1 for the simultaneous, non-simultaneous single tunnel and twin tunnels problem.
and stable state excavations of the second tunnel.

tunnel boundary, especially when = 60° 90°. The maximum com-

circumferential displacement occur at 2 = 40° in the case with = 60°. pressive stress occurs in the case with = 15° at the point with 1 = 0° ,
Overall, the displacements along two tunnel boundaries are both small while the maximum tensile stress occurs in the case with = 60° at the
when the tunnels are horizontally arranged. point with 1 = 260° . Fig. 16(b) shows that the maximum stress con-
Fig. 16 presents the hoop stresses versus 1 or 2 along tunnel centration is greater along the second tunnel boundary than that along
boundaries for the five cases. It is seen from Fig. 16(a) that the varia- the first tunnel boundary. The maximum compressive stress along the
tions of hoop stress in the five cases are much different along the first second tunnel boundary occurs in the case with = 30° and 60° , while

Fig. 13. Normalized displacements along tunnels boundaries versus 1 or 2 for simultaneous, non-simultaneous and stable excavation of the second tunnel: (a) and
(b) displacements along the first and second tunnels boundaries, respectively.

11
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

Fig. 15. Normalized displacements versus 1or 2 for various locations of the second tunnel in stable state excavation: (a) and (b) the radial and circumferential
displacements along the first tunnel boundary, respectively; (c) and (d) the radial and circumferential displacements along the second tunnel boundary, respectively.

the maximum tensile stress occurs in the case with = 15° . The fig- arrangement angles are plotted in Fig. 17. Because of the shallower
ures also indicate that the horizontal arrangement of two tunnels leads buried depth of the second tunnel with = 15° , the settlement in this
to relative smaller hoop stresses. case is obviously larger than those in other cases, while the settlement is
The ground surface settlements in the cases with different much smaller when the second tunnel is located blow the first tunnel

12
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

Fig. 16. Normalized hoop stresses in stable state excavation versus 1or 2 for various locations of the second tunnel: (a) and (b) the hoop stresses at the first and
second tunnel boundaries, respectively.

over the tunnel center when the two tunnels are vertically arranged
( = 90° ), instead of a shape of “V” with flat bottom in the other cases (it
shows as a flat bottom because of the small tunnel spacing). It is
probably because the presence of first tunnel significantly reduce the
vertical ground settlement induced by the second tunnel excavation.

6.3. Influence of tunnel spacing and the relative size of the second tunnel

In this sub-section, the buried depth of tunnels is

assumedH1 = H2 = 3R1, and the stable state excavation of second tunnel
is adopted.
Fig. 18 presents the final displacements at the given points versus
the tunnel spacing. It is noted that the absolute value of displacements
all decrease gradually with the increase of tunnel spacing, and a similar
decreasing rate can be observed. The horizontal displacements at points
with 1 = 90° and 2 = 90° tend to zero when c R1 larger than 2.5, which
indicates that the interaction between tunnels can be neglected. Fig. 19
presents the hoop stresses at the given points versus the tunnel spacing.
Compared with the variation of displacement in Fig. 18, the stresses
Fig. 17. Final surface settlement in stable excavation for various locations of also decrease with the increase of tunnel spacing, however, the de-
the second tunnel. creasing rate is much larger than that of displacement. The stresses are
almost unchanged with spacing when the spacing is larger than 2.0R1.
In order to investigate the influence of relative size of second tunnel,
the displacements versus the relative radius of second tunnel are plotted
( = 90° ). It can be also observed that the position corresponding to
in Fig. 20. The displacements nonlinearly increase with the increase of
maximum settlement gradually moves to left with the arrangement
radius of second tunnel. The greatest increasing rate of horizontal dis-
angle change from 15° to 90° . It is interesting to see that the variation
placement appears at first tunnel boundary with 1 = 0° , while that of
of settlement against x1 is taken on a shape of “W” with a slight heave
vertical displacement appears at second tunnel boundary with 2 = 90°.

13
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

Fig. 18. Normalized final displacements versus tunnel spacing at points with 1 = 0° 90° , 2 = 180° 90° along tunnels boundaries, and points O1 and O2: (a) and (b) the
horizontal and vertical displacements, respectively.

unchanged. The increase of hoop stresses at point with 2 = 90° is the

most obvious.

7. Conclusions

In this study, the analytical approaches were provided for obtaining

the time-dependent stress and displacement induced by excavation of
new tunnels in close proximity to the existing ones in semi-infinite
viscoelastic ground, by complex variable theory and the methodology
for solving a generalized viscoelastic problem. Two types of initial
stress field due to the gravity were introduced. A close agreement was
shown between the analytical solutions and numerical results under the
same simplification. The applicability of analytical solution in real en-
gineering was validated through the comparison between analytical
and FDM results under complicated geometry and ground conditions.
Finally, the parametric investigation was performed to investigate the
influence of some core parameters. From the investigation, the fol-
lowing conclusions are drawn:
Fig. 19. Normalized hoop stresses versus tunnel spacing at points with
1 = 0 90 and 2 = 180 90 along tunnels boundaries.
° ° ° °
(a) The variation patterns of surface settlement predicted by analytical
model of Case 2 (without gravity variation across the height of
excavation) are close consistent with the FDM results for real en-
It is also emerged from Fig. 20(b) that the vertical displacements at gineering. However, the variation pattern and magnitude of set-
1 = 90 and O1 are almost unchanged with the increase of the second tlements obtained by analytical model of Case 1 (with considering
°

tunnel radius when R2 R1 is larger than 1.5. The transition from nega- the gravity gradient) are both very different from FDM results.
tive to positive value for horizontal displacements at point with (b) The additional displacements around the first tunnel, which is an
2 = 180 is observed. The horizontal displacements along first tunnel exponential function of t2 (exp( Ct2 )), is significantly influenced by
°

boundary and vertical displacements along the second tunnel boundary the excavation time of the second tunnel. The final displacements
both increase obviously with the increase of the second tunnel radius. around the second tunnel are almost unaffected by the excavation
Fig. 21 presents the hoop stress at the given points versus the re- time of the second tunnel.
lative second tunnel radius. It is noted that the absolute value of hoop (c) The displacements and hoop stresses along two tunnel boundaries
stresses increase with the increase of second tunnel radius at the second are both relatively small when the tunnels are horizontally ar-
tunnel boundary with 2 = 180° 90° , whereas it decreases at the first ranged. When the tunnels are vertically arranged, the presence of
tunnel boundary ( 1 = 90° ). The stresses at point with 1 = 0° are almost the first tunnel significantly reduce the vertical ground settlement

14
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

Fig. 20. Normalized displacements versus the relative radius of second tunnel, at points with 1 = 0° 90° , 2 = 180° 90° along tunnels boundaries, and points O1 and O2 :
(a) and (b) the horizontal and vertical displacements, respectively.

The analytical solution proposed in this study can be used as an

alternative approach for the preliminary design of future shallow tun-
nels. However, the solution is under the simplifying assumption of small
deformation and unsupported condition, due to the difficulty in math-
ematical and mechanical modelling if complex conditions are con-
sidered. In the future study, the excavation of multi tunnels and the
effect of support deserve attraction.

CRediT authorship contribution statement

H.N. Wang: Conceptualization, Methodology. X. Gao:

Methodology, Writing - review & editing. L. Wu: Software, Validation,
Writing - original draft. M.J. Jiang: Supervision, Project administra-
tion.

Declaration of Competing Interest

The authors declare that they have no known competing financial

interests or personal relationships that could have appeared to influ-
ence the work reported in this paper.
Fig. 21. Normalized hoop stresses versus the relative radius of the second
tunnel, at pints with 1 = 0° 90° and 2 = 180° 90° along tunnels boundaries.
Acknowledgements

induced by the second tunnel excavation. This study was supported by the National Natural Science
(d) When the distance from center to center is larger than 2.5R1 Foundation of China (Grant Nos. 11572228, 11872281, 51639008); the
(2.0R1), the interaction between two tunnels can be neglected from State Key Laboratory of Disaster Reduction in Civil Engineering
perspective of displacement (stress). (SLDRCE19-A-06). These supports were greatly appreciated.

Appendix A:

Determination of elastic potentials for the first tunnel excavation stage [27,39]

In the first tunnel excavation stage, only one circular tunnel (the first tunnel) contains in the ground. The region of a half-plane containing a
circular hole can be mapped conformably onto an annulus in the ξ-plane, with the outer and inner radii of 1 and , respectively [34]. The mapping
equation can be expressed as follows:

15
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

1 2 1+
z= ( )= ih
1+ 2 1 (A1)
h R02
where = R0
1 1 h2
; and h is the depth of the tunnel center, = + i = ei . The following presents the outline of the derivation of

potentials for single tunnel problem, subject to tractions (1 0)

x and (1 0)
y along tunnel boundary.
The two potentials in complex variable theory for generalized shallow tunnel problem can be expressed as follows [39]:

( 1) X + iY k k
(z ) = 2 (1 + )
[ ln(z z¯0 ) + ln(z z 0 )] + ak + bk
k=0 k=1

( 1) X iY k k
(z ) = [ln(z z¯0 ) + ln(z z 0 )] + ck + dk
2 (1 + )
k=0 k=1 (A2)
1 2
where (X , Y ) represents the resultant vector of the tractions acting on the tunnel boundary; z 0 = ih 1 + 2 ; = 3 4 for plane strain problem with
being the Poisson's ratio; the coefficients ak , bk , ck , and dk are determined according to the stress boundary conditions.
According to the real engineering practice, two cases with different initial stress states are considered:
Case 1: the variation of vertical gravity loading across the height of excavation is considered in the initial stress state;
Case 2: the variation of vertical gravity loading across the height of excavation is neglected. The initial stresses along anticipated tunnel boundary
are assumed approximately equal to the stresses at the center of the tunnel [34].
In Eq. (A2), X = 0 and Y = R2 in Case 1, whereas X = Y = 0 in Case 2.
The boundary conditions, in terms of the potentials for the ground surface (zero stress), are as follows:

( ) ( 1)¯ ( , t)
( 1) ( , t) + + ( 1)¯( , t) =0
¯( )
= ei (A3)
The stress condition on the tunnel boundary is as follows:

( ) ( 1)¯ ( , t)
(1 ei ) ( 1) ( , t) + + ( 1)¯( , t) = (1 ei ) i ( b
+i b
¯( ) x y )ds + C3 ;
= e i
(A4)
where C3 is a constant, b
x and b
y are projections along the x and y of boundary stresses, respectively. b
x = K 0 y cos , b
y = y sin in Case 1, and
dy dx
b
x = K 0 h ds , yb = h ds in Case 2, where K 0 is the lateral pressure coefficient and is the angle between the outward normal direction of the tunnel
boundary and the x-axis.
According to the boundary conditions, the linear system of the equations with regard to the coefficients in the potentials can be finally de-
termined [39].

Appendix B:

Determination of elastic potentials for twin tunnel model [27]

Fig. A1 presents the alternating iterative processes for solving the problem of semi-infinite ground containing the two tunnels loaded by tractions
x and
(2 1) (2 1)
y along the second tunnel boundary (Fig. A1a) ( 1 = 2 = 1.0 ). Analysis for single-tunnel problems subject to redundant surface
tractions is repeated in the alternating iterations to eliminate the non-zero tractions on the tunnel boundaries.
In the first iteration, as illustrated in Fig. A1(b), only the second tunnel contained in the half plane is subject to tractions (2 1)
x and (2 1)
y
along the second tunnel boundary. By the derivation in Appendix A, the potentials 11 ( ) and 11 ( ) can be addressed. Subsequently, the non-zero
redundant tractions S 1 1 (Sx1 1 and Sy1 1 are the horizontal and vertical components), induced along the first tunnel boundary will be determined by

Fig. A1. Schwartz alternating iterative steps for solving the problem of semi-infinite ground containing the two tunnels loaded by tractions (2 1)
x and (2 1)
y along
the second tunnel boundary ( 1 = 2 = 1.0 ).

16
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

the following equation.

zb1 1 (z b1) d 11¯ ( ) 1¯

f1 (z b1) = i [Sx1 1
(z b1) + iSy1 1
(z b1)] ds = 1
1 (z b1) + + 1 (z b1)
zb01 ¯
1 (z b1) d
= z b1 (A5)
where and z b1 represent the certain and generic points on boundary of the first tunnel in mapping planes; 1 (z ) is the mapping function of the
0
z b1
problem only the first tunnel containing in the ground.
In order to eliminate the redundant tractions to satisfy the stress boundary condition in Fig. A1(a), the ground only containing the first tunnel is
then investigated (see Fig. A1c), where the reverse redundant tractions S 1 1 (z b1) are exerted on tunnel boundary. Therefore, the boundary
condition yields as follows:

1 1 (z b1) d 21¯ ( ) 1¯
2 (z b1) +
¯
+ 2 (z b1) = f1 (z b1)
1 (z b1) d
= zb1 (A6)
In order to easily obtain the analytical solutions through Cauchy’s Integral Formulas in the next iteration, the functions f1 (z b1) is expanded as
Fourier series. The potentials 21 ( j ) and 21 ( j ) are addressed via the procedure in Appendix A. Through the follow-up repeating iterations, the
redundant tractions along tunnel boundaries will approach zero with the increase of the iterations.
Assuming that ( 2) and ( 2) are the potentials of the twin-tunnel problem in Fig. 2(b), they are finally derived by superposing the potentials in
all iteration steps, as the following:
n
( 2) k k
( )= [ 1 ( )+ 2 ( )],
k=1
n
( 2) k k
( )= [ 1 ( )+ 2 ( )].
k=1 (A7)
where n is the final iteration number; k
j ( ) and k
j ( ) denote the potentials in the case of half-plane medium only containing the j-th tunnel in the k-
th iteration.

References [17] Hoang SK, Abousleiman YN. Extended Green's solution for the stresses in an infinite
plate with two equal or unequal circular holes. J Appl Mech 2008;75(3).
[18] Radi E. Path-independent integrals around two circular holes in an infinite plate
[1] Chang CT, Sun CW, Duann SW, Hwang RN. Response of a Taipei Rapid Transit under biaxial loading conditions. Int J Eng Sci 2011;49(9):893–914.
System (TRTS) tunnel to adjacent excavation. Tunnel Undergr Space Technol [19] Spencer SE, Sinclair GB. Stress analysis of an elastic half-space perforated by twin
Incorporat Trenchless Technol Res 2001;16(3):151–8. circular holes and under gravitational loading. Int J Mech Sci 1982;24(1):27–35.
[2] Chakeri H, Ozcelik Y, Unver B. Investigation of ground surface settlement in twin [20] Manh HT, Sulem J, Subrin D. Interaction of circular tunnels in anisotropic elastic
tunnels driven with EPBM in urban area. Arab J Geosci 2015;8(9):7655–66. ground. Géotechnique 2015;2(4):287–95.
[3] Liang RZ, Xia TD, Hong Y, Yu F. Effects of above-crossing tunnelling on the existing [21] Guo ZH, Liu XR, Zhu ZY. Elastic solution for a deep twin tunnel’s stress based on
shield tunnels. Tunn Undergr Space Technol 2016;58:159–76. complex variable theory and the superpostion principle. J Eng Res
[4] Sulem J, Panet M, Guenot A. Closure analysis in deep tunnels. Int J Rock Mech Min 2017;5(2):68–86.
Sci Geomech Abstracts 1987;24(3):145–54. [22] Ukadgaonker VG, Patil DB. Stress analysis of a plate containing two elliptical holes
[5] Wang HN, Utili S, Jiang MJ, He P. Analytical solutions for tunnels of elliptical cross- subjected to uniform pressures and tangential stresses on hole boundaries. J Eng
section in rheological rock accounting for sequential excavation. Rock Mech Rock Indust 1993;115(1):93–101.
Eng 2015;48(5):1997–2029. [23] Zhang LQ, Yue ZQ, Lee CF, Tham LG, Yang ZF. Stress solution of multiple elliptic
[6] Chehade FH, Shahrour I. Numerical analysis of the interaction between twin-tun- hole problem in plane elasticity. J Eng Mech 2003;129(12):1394–407.
nels: influence of the relative position and construction procedure. Tunn Undergr [24] Zhang LQ, Lu AZ, Yue ZQ, Yang ZF. An efficient and accurate iterative stress so-
Space Technol 2008;23(2):210–4. lution for an infinite elastic plate around two elliptic holes. subjected to uniform
[7] Liang QG, Li J, Li DW, Qu EF. Effect of blast-induced vibration from new railway loads on the hole boundaries and at infinity. Eur J Mech - A/Solids
tunnel on existing adjacent railway tunnel in Xinjiang, China. Rock Mech Rock Eng 2009;28(1):189–93.
2013;46(1):19–39. [25] Fu J, Yang JS, Yan L, Abbas SM. An analytical solution for deforming twin-parallel
[8] Do NA, Dias D, Oreste P. 3D numerical investigation on the interaction between tunnels in an elastic half plane. Int J Numer Anal Meth Geomech
mechanized twin tunnels in soft ground. Environ Earth Sci 2015;73(5):2101–13. 2014;39(5):524–38.
[9] Das R, Singh PK, Kainthola A, Panthee S, Singh TN. Numerical analysis of surface [26] Wang HN, Zeng GS, Utili S, Jiang MJ, Wu L. Analtycial soltuions of stresses and
subsidence in asymmetric parallel highway tunnels. J Rock Mech Geotech Eng displacements for deeply buried twin tunnels in viscoelastic rock. Int J Rock Mech
2017;9:170–9. Min Sci 2017;93:13–29.
[10] Dang VK, Dias D, Do NA, Vo TH. Impact of blasting at tunnel face on an existing [27] Wang HN, Wu L, Jiang MJ, Song F. Analytical stress and displacement due to twin
adjacent tunnel. Int J Geomate 2018;15(47):22–31. tunneling in an elastic semi-infinite ground subjected to surcharge loads. Int J
[11] Peck RB. Deep excavations and tunnelling in soft ground. In: Proc 7th int conf on Numer Anal Meth Geomech 2018;42(6):809–28.
SMFE; 1969. [28] Jaeger JC, Cook NGW, Zimmerman RW. Fundamentals of rock mechanics. 4th ed.
[12] Mair RJ, Taylor RN, Burland JB. Prediction of ground movements and assessment of Oxford: Blackwell Publishling; 2007.
risk of building damage due to bored tunneling. Geotech Aspects Undergr Constr [29] Unlu T, Gercek H. Effect of Poisson's ratio on the normalized radial displacements
Soft Ground; 1996. occurring around the face of a circular tunnel. Tunn Undergr Space Technol
[13] Suwansawat S, Einstein HH. Describing settlement troughs over twin tunnels using 2003;18(5):547–53.
a superposition technique. J Geotech Geoenviron Eng 2007;133(4):445–68. [30] Vlachopoulos N, Diederichs MS. Improved longitudinal displacement profiles for
[14] Ling CB. On the stresses in a plate containing two circular holes. J Appl Phys convergence confinement analysis of deep tunnels. Rock Mech Rock Eng
1948;19(1):77–82. 2009;42:131–46.
[15] Zimmerman RW. Second-Order approximation for the compression of an elastic [31] Paraskevopoulou C, Diederichs MS. Analysis of time-dependent deformation in
plate containing a pair of circular holes. ZAMM-J Appl Math Mech: Zeitschrift für tunnels using the Convergnece-Confinement Method[J]. Tunn Undergr Space
Angewandte Mathematik und Mechanik 1988;68(11):575–7. Technol 2017;71:62–80.
[16] Kooi CB, Verruijt A. Interaction of circular holes in an infinite elastic medium. Tunn [32] Muskhelishvili NI. Some basic problems of the mathematical theory of elasticity,
Undergr Space Technol 2001;16(1):59–62. 2nd ed. Groningen: Noordhoff International Publishing; 1977.

17
H.N. Wang, et al. Computers and Geotechnics 120 (2020) 103385

[33] Wang HN, Chen XP, Jiang MJ, Song F, Wu L. The analytical predictions on dis- [36] Wang HN, Zeng GS, Jiang MJ. Analytical stress and displacement around non-cir-
placement and stress around shallow tunnels subjected to surcharge loadings. Tunn cular tunnels in semi-infinite ground. Appl Math Model 2018;63:303–28.
Undergr Space Technol 2018;71:403–27. [37] Hasanpour R, Chakeri H, Ozcelik Y, Denek H. Evaluation of surface settlements in
[34] Verruijt A. Deformations of an elastic half plane with a circular cavity. Int J Solids the Istanbul metro in terms of analytical, numerical and direct measurements. Bull
Struct 1998;35(21):2795–804. Eng Geol Environ 2012;71(3):499–510.
[35] Song F, Wang HN, Jiang MJ. Analytical solutions for lined circular tunnels in vis- [38] Cristescu ND. Rock rheology. Dordrecht: Kluwer Academic Publishers; 1989.
coelastic rock considering various interface conditions. Appl Math Model [39] Lu AZ, Zeng XT, Xu Z. Solution for a circular cavity in an elastic half plane under
2018;55:109–30. gravity and arbitrary lateral stress. Int J Rock Mech Min Sci 2016;89:34–42.

18