You are on page 1of 18

Computers and Geotechnics 147 (2022) 104757

Contents lists available at ScienceDirect

Computers and Geotechnics


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

Timoshenko-beam-based response of existing tunnel to single tunneling


underneath and numerical verification of opening and dislocation
Jian Yu a, b, Hao Li a, b, Maosong Huang a, b, *, Yonghui Li c, Jorgin Qi Wen Tan d, Yuancheng Guo c
a
Department of Geotechnical Engineering, Tongji University, Shanghai 200092, China
b
Key Laboratory of Geotechnical and Underground Engineering of Ministry of Education, Tongji University, Shanghai 200092, China
c
School of Civil Engineering, Zhengzhou University, Zhengzhou 450001, China
d
Land Transport Authority, Engineer, China

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

Keywords: Longitudinal deformation of shield tunnels consists of a combination of both bending and shearing, which is
Shield tunnel characterized by the opening of the joints and the dislocation between rings. The response of existing tunnels
Timoshenko beam induced by undercrossed tunnels is thus studied with elastic continuum and Winkler solutions, considering the
Elastic continuum solution
tunnel as a Timoshenko beam with equivalent bending and shear stiffness. A displacement-based single variable
Winkler model
Numerical simulation
governing differential equation is then proposed to resolve the shear locking problem of the Timoshenko beam.
The applicability of the proposed Winkler modulus is verified against the elastic continuum solution. Next, the
applicability of the equivalent stiffnesses in the prediction of the opening and dislocation is examined using a
field case and an elaborate finite element shield tunnel model that considers the soil as the PSI (pipe/beam-soil
interaction) element with the proposed Winkler spring stiffness. Finally, the Winkler model is used to establish
four design charts for quick assessments of the maximum opening of the joints and the maximum dislocation
between rings, and the maximum tunnel displacement. Their effectiveness is verified by compared with the finite
element results.

1. Introduction the circumferential joints. Similar investigations have been also reported
by Chen et al. (2021). Therefore, to protect the existing shield tunnels,
The rapid developments of modern cities and population growth strict deformation control standards have been established. The limiting
have led to a severe shortage of land, especially in areas with dense values of the additional opening and dislocation are recommended as 2
populations. These developments were often associated with an mm and 4 mm by the standard for design of shield tunnel engineering
increasing demand for public transport, leading to many tunnels being (GB/T 51438-2021 Standard for design of shield tunnel engineering,
constructed. Such constructions would generate ground movement 2021) and the code for design of Zhejiang province urban rail transit
(Peck, 1969; Mair et al., 1993; Vorster et al., 2005), which could (DB33/T 1146-2018 Code for design of Zhejiang province urban rail
potentially risk the structural stability or serviceability of existing transit, 2018), respectively. The codes also require to strengthen the
structures. Due to the existence of segment joints, the overall stiffness of monitoring of the structure and take protective measures once the
a shield tunnel is reduced. The longitudinal deformation of a shield monitored values reach 60% and 80% of the limiting values, respec­
tunnel is a combination of both bending and shearing, characterized by tively. Thus, it is worth a further study on the effect of tunneling on an
the opening of the joints and the dislocation between rings (Shen et al., existing nearby tunnel.
2014; Gue et al., 2015). This mode of deformation makes the shield To study the influence of joints on shield tunnels, elaborate numer­
tunnel very sensitive to the changes in the surrounding environment. ical simulation methods are typically adopted (Wang et al., 2014b; Shi
Liu et al. (2020) reported a case study on the structural responses of et al., 2016b; Chen et al., 2020). Wang et al. (2014b) simulated the
shield tunnels caused by unexpected soil and water loss at the bottom of failure mechanism of tunnel segmental lining joints and longitudinal
the tunnel. Severe opening and dislocation occurred between the adja­ bolts under the uneven longitudinal ground settlement. In their study, a
cent rings, further causing serious concrete cracking and leakage around numerical model consisting of a tunnel with transversely homogeneous

* Corresponding author.
E-mail address: mshuang@tongji.edu.cn (M. Huang).

https://doi.org/10.1016/j.compgeo.2022.104757
Received 21 December 2021; Received in revised form 3 April 2022; Accepted 7 April 2022
Available online 21 April 2022
0266-352X/© 2022 Elsevier Ltd. All rights reserved.
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

lining rings and 3D bolts was adopted and the ground reactions were
modeled as ground springs located along the tunnel periphery. Shi et al.
(2016b) introduced a multiscale mixed modeling technique to study the φ
effects on the tunnel segment joints due to lateral unloading from the
tunnel excavation. In the local area significantly affected by excavation, O x
the joint, bolt, and rubber sealing gasket between segments are wb
considered. The other parts are handled as a homogeneous tube model. w ws dw
Chen et al. (2020) established an elaborate numerical model incorpo­
rating reinforced segments and segment bolts to investigate the struc­ dx
tural behavior of the segmental ring. An elaborate numerical simulation
method can better simulate the mechanical properties of tunnel joints,
especially when demonstrating the opening and dislocation of joints.
γ
However, numerical modeling and calculation are generally time-
consuming and require a certain level of technical background and
expertise.
Many semi-analytical methods have been developed to estimate the y
responses of existing tunnels induced by new tunnels. Two-stage-based
Fig.1. Timoshenko beam element deformation analysis.
elastic continuum methods were proposed by (Klar et al., 2005; Zhang
et al., 2012), in which the tunnel and soil are treated as an Euler beam
and an elastic continuum respectively. The two stages consist of first of the joints and the maximum dislocation between rings, and the
estimating the tunnel-induced green soil movement at the existing maximum tunnel displacement, respectively. The effectiveness of the
tunnel centerline level and then calculating the responses of the existing proposed charts is then verified and compared with the finite element
tunnel when subjected to soil movement. The Euler beam was also results.
combined with the Winkler model to simulate the tunnel-soil interac­
tion, as it is easier to further consider the soil nonlinear behavior 2. Timoshenko-beam-based elastic response of an existing
(Attewell et al., 1986; Huang et al., 2009; Zhang and Huang, 2014; tunnel due to new tunneling underneath
Liang, 2019). To consider the effect of the joints, the equivalent bending
stiffness should be input (Shiba et al., 1988; Liao et al., 2008; Wu et al., 2.1. Single-variable formulation of Timoshenko beam
2015; Cheng et al., 2021). To further consider the coupling of the
shearing and bending deformation, the Timoshenko beam was also Timoshenko beam theory (Timoshenko, 1921) is suitable to simulate
adopted in the Winkler model with the Vesic modulus (Li et al., 2016; the tunnel longitudinal behaviors (Wu et al., 2015) as it accounts for
Liang et al., 2017). However, the previous studies by compared with both bending and shearing deformation of a beam. The equilibrium
numerical results and elastic solutions (Klar et al., 2005; Basmaji et al., equations, the kinematic relationships, and the material laws for a
2019) reveal that the Winkler model with the Vesic’s modulus would Timoshenko beam are expressed respectively as:
underestimate the response of the structure with an input of ground dM dV
movement, as the Vesic’s modulus used in the Winkler modulus was V= ,q = − , (1)
dx dx
obtained by allowing an infinite beam resting on the surface of the soil
(Winkler support) to exhibit similar displacements and moments to that dw dφ d2 φ dγ
φ= − γ, kc = − = − + , (2)
on an elastic half-space when loaded with the same concentrated force dx dx dx2 dx
or moment (Yu et al. 2013). Thus, for the Winkler-Timoshenko solution,
the applicability of the subgrade modulus needs to be further examined. M = EIkc , V = κGAγ, (3)
Design charts have been also developed to predict the structural
response due to ground movements. Klar et al. (2005) and Huang et al. where M is the bending moment, V is the shear force, q is the applied
(2019) developed the relationship between the normalized maximum transverse load, w is the deflection of the neutral axis of the beam, γ is
bending moments/joint rotation angle and the relative soil-pipe stiffness the shear angle, φ is the rotation angle of the cross-section, kc is the
when studying the tunnel-pipeline interaction. Goh and Mair (2014) and curvature of the neutral axis, E is Young’s modulus, I the is the area
Franza et al. (2020) established and improved the design chart for moment of inertia of the cross-section, G the is the shear modulus, A is
evaluating building damage due to tunneling or excavation. Franza and the cross-sectional area, κ is the shear coefficient, equal to 0.5 for
Viggiani (2021) demonstrated the importance of the shear factor in annular cross-sections, EI is the bending stiffness, κGA is the shear
shear and bending deformations of the affected tunnel and established a stiffness (κA can be further written as As , cross-sectional shear area).
series of design charts for estimating the induced maximum settlements By re-arranging these equations, the standard two-variable differ­
and internal forces based on the greenfield with a standard Gaussian ential equations for the Timoshenko beam are expressed as (Timo­
curve. The tunnel joint deformations (opening and dislocation) under shenko, 1921):

different greenfield conditions need a further investigation. These ⎪ d2 φ
(
dw
)

design charts are usually established by numerous parametric studies ⎪
⎨ EI dx2 + κGA dx − φ = 0

based on well-established semi-analytical methods. ( 2 ) (4)

This study attempts to establish the elastic continuum and Winkler ⎪


d w dφ
⎩ κGA − = − q
methods based on the Single-variable formulation of the Timoshenko dx2 dx
beam, which would solve the shear locking problem of the Timoshenko
To solve the shear locking problem of the Timoshenko beam, Kiendl
beam. The applicability of Yu’s et al. (2013) Winkler modulus and the
et al. (2015) derived a displacement-based single variable governing
equivalent bending and shear stiffness in the deformation response of
differential equation. As shown in Fig. 1, the displacement w of the
the existing tunnel caused by tunnel excavation (especially the opening
neutral axis of the Timoshenko beam consists of two parts: the first is the
of the joints and the dislocation between rings) are then verified against
displacement wb caused by bending moment (bending displacement);
the elastic continuum solutions and the finite-element result of two field
the second is the displacement ws caused by shear force (shearing
cases. Finally, four design charts are established by parametric studies
displacement):
with the Winkler-Timoshenko solution to assess the maximum opening

2
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

Fig. 2. Schematic of the effect of new tunnel excavation on the existing tunnel above it.

Existing Timoshenko
tunnel beam
Fictitious qi Fictitious
l nodes
nodes Greenfield soil
displacement wCAT
fi
Fig. 3. Schematic of elastic continuum method.

w = wb + ws (5) nodes in total, inclusive of four fictitious nodes). When accounting for
the tunnel-soil interaction and the single variable governing the differ­
The single variable governing the differential equation of the Tim­
ential equation of the Timoshenko beam as shown in Eq.(6a), the lon­
oshenko beam in terms of the bending displacement ws can be written
gitudinal bending behavior of the tunnel may be represented by the
as:
following equation:
d4 wb (9)
EI =q (6a) [S]{wb } = {q}
dx4
where {wb } is the bending displacement, [S] is the bending stiffness
EI d2 wb
w = wb − (6b) matrix of the tunnel, {q} is a force vector representing the soil loading
κGA dx2 acting on the beam elements. To obtain the bending displacement of the
Then, all variables of the beam model can be expressed in terms of fictitious nodes at both ends of the existing tunnel, both ends of the
the bending displacement wb as follows: tunnel are to satisfy the free boundary condition, i.e., the bending mo­
ments and shear forces in Eq. (8) are equal to zero. Thus, the stiffness
EI d2 wb dwb EI d3 wb
ws = − ,φ = ,γ = − , (7) matrixes [S] can be obtained as follows:
κGA dx 2 dx κGA dx3
⎡ ⎤(n + 1)×(n + 1)
2 − 4 2
d2 wb d3 wb ⎢− 2 5 ⎥
M = − EI 2 , V = − EI 3 , (8) ⎢ − 4 1 ⋯ 0 ⎥
dx dx ⎢ 1 − 4 6 − 4 1 ⎥
EI ⎢ ⎥
The equation proposed by Kiendl et al. (2015) has the merit of [S] = 4 ⎢ ⋮ ⋱ ⋮ ⎥ (10)
l ⎢


1 − 4 6 − 4 1 ⎥
expressing both the differential governing equation and the boundary ⎢


0 ⋯ 1 − 4 5 − 2⎦
conditions in terms of a single variable and being completely locking-
2 − 4 2
free.
Following Klar et al.’s (2005) assumption that the affected pipeline is
always in contact with the soil around it, a perfect compatibility con­
2.2. Elastic solution of an existing tunnel undercrossed by a new tunnel dition is employed in this study. However, their later experimental and
numerical studies (Vorster et al. 2005) demonstrated that the perfect-
Fig. 2 illustrates the ground settlement of an existing tunnel under­ compatibility-based linear elastic solution would overestimate the true
crossed by a new tunnel. As previously mentioned, the two-stage method pipeline response, as for a given greenfield settlement trough, any soil
is usually adopted for such situations (Attewell et al., 1986; Huang et al., nonlinearity will reduce the stiffness, thereby resulting in that the
2009), where the tunnel-induced greenfield soil movement at the pipeline is less constrained to move with the greenfield soil. Analo­
existing tunnel centerline level is first estimated and the responses of the gously, Lin et al. (2019) presented more detailed descriptions on the
existing tunnel subjected to greenfield soil movement is then calculated. development of the existing tunnel deformation behavior and the stress
The elastic continuum solution for an existing tunnel undercrossed of the surrounding soil with new tunnels construction undercrossing
by a new tunnel is given next. To simplify the calculation process, the based on a numerical simulation of a case history. It revealed that the
finite difference method is utilized to solve the problem numerically. new tunnel excavation caused an increase in the vertical stress at the
Referring to Fig. 3, the beam is divided into n elements of length l (n + 5

3
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

crown of the existing tunnel and a reduction at the bottom, respectively. Table 1
The formation of the soil arching zone at the crown further induced Parameters from back analysis of centrifuge tests.
stress redistribution. Case Smax i α
To consider the effect of soil nonlinearity illustrated above, Vorster (m) (m)
et al. (2005) deduced an expression to estimate the average shear strain Case 1 0.0089 0.286 0.439
caused by tunnel-induced ground movement. Further combined with the (Wang et al., 2014a)
soil stiffness degradation curve, the perfect-compatibility-based linear Case 2 Test 1 0.00085 0.095 0.153
elastic solution was used to predict the centrifuge test results success­ (Ma et al., 2017) Test 2 0.00084 0.097 0.167
Case 3 Test P 0.00069 0.0613 0.033
fully. The idea is also demonstrated in the building-tunnel interaction (Shi et al., 2016a) Test O 0.00078 0.0697 0.067
(Yu et al. 2022). Case 4 0.0115 6.18 0.187
To satisfy the compatibility condition, the displacement vector of the (Jia et al., 2009)
existing tunnel, {w}, is given by.
⎧ ⎧
⎪ [ *]
⎪ ⎪


⎪ {w} = {wCL } + {wCAP } + {wCAT } ⎨ {wCAP } = λs {f }

⎪ {
⎨ ∑n [ *] Gi,j , i ∕
=j (17)
{wCL }i = {f }i Gi,i and {wCAP }i = j = 1 {f }j Gi,j (11) ⎪
⎩ λs =


⎪ 0, i = j



⎪ j∕=i

Considering.

where {wCL } is defined herein as local displacement, which is the {f } = − {q} = [S][wb ] (18)
displacement at a point due solely to its loading, {wCAP } is the additional Eq. (16) becomes,
displacement due to forces resulting from soil-tunnel interaction, and ( [ ] )
{wCAT } is the greenfield displacement, {f} are the forces acting on the [S] + [K * ][ST ] + [K * ] + [K * ] λ*s [S] {wb } = [K * ]{wCAT } (19)
soil medium, Gi,j is defined as the elastic soil continuum displacement at
point i due to unit loading at point j based on Mindlin’s (1936) solution. which is solved numerically to obtain the elastic continuum solution
It should be noted that the solution does not satisfy displacement at the with the Timoshenko beam, as long as wCAT is determined. Once wb is
point of loading. Therefore, Klar et al. (2005) proposed that a reference determined, w, M and Q can be calculated using both Eqs. (6b) and (8).
displacement value for that point is considered to be the average It is also noted that when the shear stiffness κGA goes to infinity,
displacement around the circumference of the tunnel. This is identical to {wb } approaches {w} and [ST ] approaches zero matrix. As such, Eq. (19)
assume a barrel load around the tunnel. Any displacement at a point due reduces to the elastic continuum solution with the Euler beam (Klar
to a uniform load is equal to the average displacement, over the same et al., 2005).
area (or volume) as that of the uniform load, due to an equivalent The standard Gaussian curve proposed by Peck (1969) is typically
concentrated load at that point. used to describe the shape of the greenfield settlement due to tunneling.
According to Eq. (6b), However, test results from (Vorster et al., 2005) showed that the stan­
dard Gaussian curve does not always accurately fit the measured soil
{w} = ([I] + [ST ] ){wb } (12) settlements. To increase the degrees of freedom for a better fit, a
modified Gaussian curve was proposed by (Vorster et al., 2005) as
where [I] is the identity matrix, [ST ] is the finite difference coefficient follows:
matrix and the determination of [ST ] also needs to be incorporated with ⎧
the boundary conditions. ⎪
⎪ N



wCAT = Smax [ (x)2 ]
⎡ ⎤(n+1)×(n+1) ⎨ (N − 1) + exp α
0 i (20)
⎢− 1 2 − 1 ⋯ 0 ⎥ ⎪

⎢ ⎥ ⎪
⎪ 2 α − 1
⎢ ⎥ ⎪
⎩ N = eα +1
EI ⎢ ⎥ 2α + 1
[ST ] = ⎢ ⋮ ⋱ ⋮ ⎥ (13)
κGAl2 ⎢





⎥ where Smax is the maximum trough settlement; x is the horizontal offset
0 ⋯ − 1 2 − 1⎦
0 from the tunnel centerline; i is the distance from the tunnel centerline to
the point of inflection of the trough, and α allows the curvature to be
Combining Eq. (11) and (12), adjusted while maintaining i as the point of inflection. When α equals
0.5, Eq. (20) reduces to the standard Gaussian curve. Table 1 lists the
([I] + [ST ] ){wb } = {wCL } + {wCAP } + {wCAT } (14)
Smax , i and α back-calculated from the measured greenfield settlement
Because of the principle of action-reaction forces, one can get: trough from the reported literature. It is found that the values of α
usually ranges from 0.01 to 0.5.
{wCL }i
{q}i = − {f }i = (15)
Gi,i
2.3. Simplification to Winkler-Timoshenko solution
By introducing this and Eq. (15) into Eq. (14) the following relation is
obtained:
To further simplify the calculation, Eq. (19) can be also converted
[S]{wb } + [K * ][ST ]{wb } + [K * ]{wb } = [K * ]{wCAP } + [K * ]{wCAT } from the elastic continuum solution to the Winkler-Timoshenko solution
⎧ given in Eq. (21) by omitting the continuum effects between the soil

⎨ 1 ,i=j

(16) elements, where λ*si,j is set as zero and [K* ] is set as the stiffness matrixes
[K * ] = Gi,i {
⎪ 1/k, i = j

⎩ 0, i ∕ of the Winkler foundation [Ks ] with the element Ksi,j = .
=j 0, i ∕
=j

([S] + [Ks ][ST ] + [Ks ] ){wb } = [Ks ]{wCAT } (21)


where [K* ] is the local stiffness matrix of the soil.
Eq. (11) shows that. It implies that the Timoshenko beam is resting on a series of inde­
pendent Winkler springs, as shown in Fig. 4.

4
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

Existing Timoshenko
tunnel beam
l
Fictitious Fictitious
nodes nodes
Greenfield soil
displacement wCAT

Fig.4. Schematic of Winkler-Timoshenko method.

Fig. 5. Comparison of maximum settlements by elastic continuum and Winkler solution.

Klar et al. (2005) pointed out that it is always safer to use the elastic where Es is the elastic modulus of soil, D is the outer diameter of the
continuum solution rather than the Winkler solution with Vesic’s (1961) existing tunnel.
subgrade modulus when predicting the response of pipeline due to Franza and Viggiani (2021) have recently pointed out that the
tunneling, as it was derived by allowing an infinite beam resting on the importance of the shear flexibility for buried tunnels depends on both
surface of the soil (Winkler support) to exhibit similar displacements and EI
the cross-sectional stiffness ratio GA and the greenfield input shape (i),
moments to that on an elastic half-space when loaded with the same s

concentrated force or moment rather than the input of ground move­ and introduced a new dimensionless shear factor Q( = GAEIs i2 ). It was also
ments. To solve the problem, Yu et al. (2013) presented an expression of indicated that the shear factor controls whether the response of buried
the subgrade modulus k for an Euler-beam-based pipeline or tunnel infrastructure to tunneling is fully described by bending (Q < 0.26),
buried at an arbitrary depth subjected to ground movements. The shear (Q > 2.6), or mixed bending/shear dominated mode (Q =
applicability of the Timoshenko beam needs to be further examined. Yu 0.26 ∼ 2.6).
et al.’s (2013) modulus is expressed as. Therefore, Figs. 5-8 compare the normalized maximum settlements
sag hog
⎧ √̅̅̅̅̅̅̅̅̅̅ wmax , sagging bending moments Mmax , hogging bending moments Mmax


⎪ 3.08 Es 8 Es D4 and shear forces Vmax resulting from the elastic continuum solutions and
⎪ k=



⎪ η 1 − ν2 EI the Winkler-Timoshenko solutions based on four different Q
⎨ ⎧ (0.026,0.26, 2.6 and 26) and the standard Gaussian curve (α = 0.5),
⎪ (22)




⎨ 2.18 z/D⩽0.5 respectively. The results of Franza and Viggiani (2021) for i/r0 = 10 are




η =

1 also given as queen stars. The comparisons illustrate that the present

⎩ ⎩ 1 + 1.7z/D z/D > 0.5

solutions can fully capture the beam behavior varying from bending to

5
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

Fig. 6. Comparison of maximum sagging bending moments by elastic continuum and Winkler solution.

shear dominated modes when subjected to the ground movements. segment, respectively.
More importantly, they demonstrate that the Winkler-Timoshenko The equivalent longitudinal bending stiffness (EI)eq (Cheng et al.,
solutions with Yu et al.’s (2013) modulus show good agreements with 2021) is expressed as follows:
the continuum solutions for the different ratios i/r0 (r0 = D/2) and Q. ⎧
ls
By contrast, those with Vesic modulus show a significant underestima­ ⎪

⎪ (EI)eq = Ec Ic
⎪ ( π)
tion. Generally, for a shield tunnel with a buried depth, the effect of the ⎪

⎨ cosθ + θ + sinθ
2
buried depth on the subgrade modulus should be considered (Attewell ls − λlb + λlb
cos3 θ (24)

et al. 1986). However, the Vesic’s modulus is obtained by allowing the ⎪

⎪ ( )
⎪ 1 n b Eb Ab
beam resting on the ground surface. Thus, the use of Vesic’s modulus ⎪
⎩ θ + cotθ = π +
2 Ec Ac
leads to underestimated tunnel responses.
where ls is the length of the shield segment, lb is the length of longitu­
3. Verification of the equivalent stiffnesses for shield tunnels dinal steel bolts, λ is the influencing factor of circumferential joints
0.4725 (Xu, 2005), θ is the angle of the neutral axis, Ec is the elastic
3.1. Equivalent longitudinal bending stiffness and shear stiffness modulus of the concrete shield segment, Ic is the moment of inertia of the
tunnel cross-section, nb is the number of longitudinal steel bolts, Eb is the
The Timoshenko-beam-based Winkler solution is based on the lon­ elastic modulus of the bolt, Ab is the cross-sectional area of the bolt, Ac
gitudinal continuous model. However, the structure of the shield tunnel the cross-sectional area of the tunnel. Note that EI and κGA are specified
is composed of concrete rings connected by bolts, which weakens the as (EI)eq and (κGA)eq in the following text.
structure’s longitudinal stiffness between the rings. Wu et al. (2015)
Using both Eqs. (23) and (24), Table 2 summarizes the equivalent
deduced the equivalent shear stiffness, while Cheng et al. (2021)
bending stiffness and equivalent shear stiffness of common shield tun­
developed the equivalent bending stiffnesses from the work of Shiba
nels in China. (E/G)eq is the ratio of the equivalent elastic modulus to the
et al. (1988) and Liao et al. (2008). The equivalent shear stiffness
equivalent shear modulus of the Timoshenko beam, which is estimated
(κGA)eq (Wu et al., 2015) is expressed as follows:
using the equivalent stiffnesses and tunnel section geometry. It can be
ls seen that the ratio (E/G)eq is relatively similar, ranging from 6.5 to 8,
(κGA)eq = ξ (23)
which implies that the bolts at the joint location do decrease the overall
lb
nb κb Gb Ab
+ κlcsG− clAb c
shear stiffness of the tunnel structure, as the value of (E/G) is approxi­
where ξ is the modification factor used to consider the actual contact mately 2.6 for a isotropic material. Similarly, Tomaževič (2009) had
relations between shield segments and is taken as 1 in the following recommended a (E/G)eq ratio of 10 for masonry façade due to the
analysis (Liang et al., 2017), κb and κc are the shear coefficient of the bolt reduction of shear stiffness by door and window opening.
and shield segment for the Timoshenko beam theory, and are set to 0.9 The applicability of the equivalent shear and bending stiffness have
and 0.5, respectively. Gb and Gc are the shear modulus of the bolt and been demonstrated in the prediction of tunnel settlements (Li et al.,

6
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

Fig. 7. Comparison of maximum hogging bending moments by elastic continuum and Winkler solution.

2016; Liang et al., 2017; Liu et al. 2020; Cheng et al., 2021;). Liu et al. To better model the opening and the dislocation at joints, an elaborate
(2020) indicated that openings and dislocations between rings caused by tunnel numerical model with solid lining rings connecting by solid bolts
the differential settlement pose a potential risk to the tunnel’s structure, is employed in this study (Wang et al., 2014b; Shi et al., 2016b; Chen
i.e. the cracking, peeling of the ring, yielding of the joint bolts, and local et al., 2020).
leakage. Based on these equivalent stiffnesses, the opening of the joints A case study (Jin et al., 2018b) in Shenzhen where a new shield
at the edge of the ring Δ and the dislocation between adjacent rings δ can tunnel was excavated at a depth of 21 m (Shenzhen metro line 9) below
be calculated in terms of the bending moment M and shear force Q of an existing 13.5 m deep tunnel (Shenzhen metro line 3, see Fig. 9), is
continuous Winkler-Timoshenko solutions respectively as: adopted as an example in this study, in which the material properties
and geometric parameters for the lining rings and the bolts of Shenzhen
Mλlb cosθ + (π /2 + θ)sinθ
Δ= (r0 + rsinθ) (25) metro line 3 are listed in Table 3. This paper focuses on the longitudinal
Ec I c cos3 θ
mechanical behavior of shield tunnels; therefore, the tunnel lining is
V simplified as a homogeneous and continuous ring and the transverse
δ = ls tan (26) effective rigidity ratio of the lining ring is taken as 0.7 (Liao et al., 2008).
(κGA)eq
As shown in Fig. 10, each lining ring and bolt consist of 1920 and 160
where r0 is the outer radius of the existing tunnel, r is the distance from C3D8R elements, respectively. Structured meshing technique is adopted
the center of the ring to the location of the bolt. However, compared as the meshing control method. A linear elastic material model is used to
with the settlement, the opening of joints and the dislocation are more describe the mechanical behavior. The model of the joints between the
sensitive to the accuracy of the equivalent stiffnesses. Therefore, it re­ lining rings is shown in Fig. 10(a). Each lining circumferential joint is
quires further evaluation by conducting an elaborate numerical simu­ connected by 10 longitudinal M24 bolts. In the model, the nuts of the
lation for the tunnel structure. bolts are embedded in the lining ring. Two surface-to-surface contact
pairs are adopted to model the interactions between 1) two rings and 2)
the bolt and the bolt hole, respectively. The normal behavior is a hard
3.2. Numerical method with PSI elements contact that allows separation, and the tangential behavior is described
by a penalty function model with a friction coefficient of 0.5.
To reduce computational costs, tunnel structures were convention­ Wang et al. (2014b), Shi et al. (2016b) and Chen et al. (2020)
ally simplified as continuous beam elements or shell elements by employed the spring elements along the tunnel periphery to model the
inputting equivalent model parameters to obtain a reasonable longitu­ soil-tunnel interaction and study the tunnel longitudinal and transverse
dinal response (Li et al., 2018; Huang et al., 2020). Liu et al. (2021) responses. Considering that this study focuses on the longitudinal re­
established a tunnel structure model by connecting solid lining rings sponses, the soil–tunnel interaction will be simulated using PSI elements
with equivalent springs, whose stiffness value was determined following along the tunnel for simplification, which has been used to model the
empirical equations. However, these simplified numerical models are interaction between a pipe and the surrounding soil (Wham et al., 2016).
unsuitable for examining the applicability of the equivalent stiffnesses.

7
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

1.0 1.0
Style Color Style Color
Yu modulus Q=0.026 Yu modulus Q=0.026
Vesic modulus Q=0.26 0.8 Vesic modulus Q=0.26
0.8
Continuum solution Q=2.6 Continuum solution Q=2.6
Q=26 Q=26

Vmaxi/GAsSmax
Vmaxi/GAsSmax
0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
10-1 100 101 102 103 104 105 10-1 100 101 102 103 104 105
EI/Esr04 EI/Esr04
(a) i/r0 = 2.5 (b) i/r0 = 5

1.0 Style Color


Yu modulus Q=0.026
Vesic modulus Q=0.26
0.8 Continuum solution Q=2.6
Franza and Viggiani (2021) Q=26
Vmaxi/GAsSmax

0.6

0.4

0.2

0.0
10-1 100 101 102 103 104 105
EI/Esr04
(c) i/r0 = 10

Fig. 8. Comparison of maximum shear forces by elastic continuum and Winkler solution.

Table 2
Equivalent bending stiffness and shear stiffness of common shield tunnels.
Tunnel Outer Lining thickness Lining Number of Diameter Length of (EI)eq (κGA)eq (E/
Diameter (m) width bolts of bolt bolt (kN⋅m2) (kN) G)eq
(m) (m) (mm) (mm)

Shenzhen Metro Line 3&4 6 0.3 1.5 10 24 400 7.78 × 107 1.18 × 106 8.08
(Jin et al., 2018a, 2018b)
Shanghai Metro Line 9 6.2 0.35 1.2 17 30 400 1.58 × 108 2.48 × 106 7.43
Typical shield tunnel in 6.2 0.35 1 17 30 400 1.36 × 108 2.08 × 106 7.57
Tianjin
Shenzhen Metro Line 11 6.7 0.35 1.5 17 30 400 2.27 × 108 3.06 × 106 7.34
Chengdu Metro Line 17 8.3 0.4 1.5 19 30 400 4.12 × 108 3.46 × 106 7.62
Beijing Metro Line 14 10 0.5 1.8 36 36 400 1.63 × 109 1.08 × 107 6.65
Shanghai Yan’an road tunnel 11 0.55 1 32 38 760 7.31 × 108 3.38 × 106 7.89

Table 3
Material properties and geometric parameters for the tunnel structure of
Shenzhen Metro Line 3.
Lining ring Ec 3.45 × 104 MPa
v 0.2
D 6m
t 300 mm
Bolt Eb 2.06 × 105 MPa
v 0.3
ls 1.5 m
lb 0.4 m
Db 24 mm
nb 10
Fig. 9. Layout of existing tunnel and new tunnel.
Note: t = Thickness of existing tunnel; Db = Diameter of bolt.

8
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

Screw

Nut

(a) Segment ring (b) Bolt

Fig. 10. 3D finite element model of the tunnel structure.

Beam
element

Coupling
constraint Solid
element

(a) Coupling constraint

Tunnel (Solid element)


Beam element

PSI element

Greenfeild soil
displacement wCAT

(b) Overall diagram of the model


Fig. 11. Tunnel–soil interaction model with PSI element.

However, it should be noted that the PSI element cannot be used directly 1,200,464 elements) are included in the FE model (see Fig. 11(b)).
with the solid element. Therefore, beam elements (B31) with a very low
stiffness need to be created at the centreline of the tunnel to connect one
3.3. Comparison between numerical and Winkler-Timoshenko solutions
end of the PSI element (PSI34) (see Fig. 11(a)). Thereafter, through the
coupling constraint command, the beam elements and the solid elements
For the tunnel structure of Shenzhen Metro Line 3, the equivalent
of the shield tunnel are constrained together to allow for common
longitudinal bending stiffness and shear stiffness are calculated as 7.78
deformation and establish the tunnel-soil interaction model as shown in
× 107 kN⋅m2 and 1.18 × 106 kN in terms of Eqs. (23) and (24).
Fig. 11(b). According to the monitored ground settlements induced by
Fig. 12 shows the comparison of tunnel settlements obtained from
Shenzhen Metro Line 9 (Jin et al., 2018b), the greenfield at the level
the FE simulation, the Winkler-Timoshenko solution, and the measured
tunnel axis has the maximum settlement Smax of around 4.39 mm, the
data. The FE result is approximately consistent with the measured data,
distance inflection points i of 5.83 m and the curvature parameter α of
which implies that the present elaborate finite element model (FEM) can
0.5. The newly built tunnel caused a volume loss of 0.37%. The obtained
provide a reasonable evaluation for the weakness of the tunnel structure
Gaussian curve is then prescribed to the other end of the PSI element as
stiffness at the joints. It also demonstrates the applicability of the
the displacement boundary conditions (see Fig. 11(b)). The stiffness of
equivalent stiffnesses for the Winkler-Timoshenko solution.
the PSI element is calculated by Yu et al’s (2013) expression given in Eq.
Fig. 13 shows the longitudinal distributions of the openings of joints
(22), where the reported soil elastic modulus of 20.2 MPa is used. To
and dislocation between the rings obtained from the numerical simu­
eliminate the boundary effect, a total of 200 rings (300 m in length,
lation and the Winkler-Timoshenko solution. Unlike the Winkler-

9
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

x/m predict relatively larger maximum dislocations than those of the FEM.
-50 -25 0 25 50 With the increase in the soil modulus, the FEM results become relatively
-1 larger. The Winkler-Timoshenko solutions generally have a consistent
trend with the FEM results. Figs. 5-8 and Figs. 15-17 show that the
0 Winkler solutions with Yu et al.’s (2013) modulus can provide similar
maximum values and also similar profiles to the continuum solutions.
Displacement/mm

1 Fig. 18 shows the effects of the shear factor Q on the normalized


sagging bending moments, hogging bending moments and shearing
2 forces. The data points of the four cases lie between two curves corre­
sponding respectively to the bending-dominated (Q < 0.26) and shear-
3
dominated (Q > 2.6) modes, which implies that the tunnel deforma­
4 tion belongs to the mixed bending/shear mode. As Shenzhen Metro Line
FEM 3 has a shear factor Q of 1.95, the deformation mode is closer to the
5 Winkler solution shear-dominated behaviour.
Measured tunnel settlement
Greenfield settlement
6
3.4. Analysis of field case
Fig. 12. Calculated tunnel settlements versus measurement and numeri­
cal result. Fig. 19 shows a case study (Fan, 2008) in Shanghai with a new shield
tunnel undercrossing at a depth of 31.462 m (South Xizang Road Tun­
Timoshenko solution’s overestimated settlement, a slight underestima­ nel) below an existing tunnel buried at a depth of 19.352 m (Shanghai
tion is observed for the openings of joints. This is mainly due to the metro line 8). The calculation parameters of the soil and the tunnel
openings of joints and dislocation being calculated indirectly using Eqs. structure are shown in Table 4. The volume loss Vs induced by the new
(25) and (26), unlike the FE results where deformation of the rings can tunnel is 0.625%. According to the study of Lin et al. (2020), the
be directly measured. The numerical simulation can better reflect the greenfield at the existing tunnel axis level has a distance inflection point
local weakening effect of the joint on the tunnel structure, while the i of 6.67 m, the maximum settlement Smax of around 39.3 mm, and the
Winkler-Timoshenko solution using the equivalent beam model gener­ curvature parameter α of 0.5. Combining Vorster et al.’s (2005)
alizes this weakening effect to the whole structure. expression for estimating the soil average strain around the tunnel and
It is worth noting that the opening calculated from the Winkler- the small-strain shear modulus Gmax and the stiffness degradation curve
Timoshenko solution shows a smooth transition from negative values of the Shanghai soil reported by Gu et al. (2021), the secant modulus of
(opening occurring at the crown) to positive values (opening occurring the soil around the tunnel is set to 35 MPa. For the tunnel structure of
at the invert), as it is based on the theory of continuous beam. However, Shanghai Metro Line 8, the equivalent longitudinal bending stiffness and
as shown in Fig. 14, for a shield tunnel consisting of rings, a discontin­ shear stiffness are calculated as1.58 × 108 kN⋅m2 and 2.68 × 106 kN in
uous transition can be seen between sagging and hogging zones. terms of Eqs. (23) and (24).
A parametric investigation is conducted to examine the applicability Fig. 20 presents the deformation of the existing shield tunnel, and
of the equivalent stiffnesses under different tunnel-soil relative stiffness obvious openings and dislocations appearing at the tunnel joints. With a
by varying the soil elastic modulus (10 MPa, 20.2 MPa, 65 MPa, and 100 shear factor Q of 1.33, a mixed bending/shear deformation mode is
MPa). Fig. 14 presents the deformation of the existing shield tunnel achieved (as shown in Fig. 18). Figs. 21-22 show the comparison of the
under four soil conditions. It is found that the settlements of the tunnel, tunnel deformation obtained from the numerical simulation, the Win­
the opening of the joints, and the dislocations between the rings grad­ kler solution and measurement data. Same as aforementioned, the
ually increase as soil stiffness is increased. Figs. 15-17 show the tunnel Winkler solutions have a consistent trend with the FEM results. The FEM
displacements, the dislocation between rings, and opening of joints results show that the maximum dislocation and opening are 1.47 mm
computed by the FEM and the Winkler-Timoshenko solution and Con­ and 1.3 mm, respectively. According to the standard for design of shield
tinuum solution, respectively. The maximum tunnel displacements of tunnel engineering (GB/T 51438-2021 Standard for design of shield
the Winkler-Timoshenko solutions are slightly larger than those of the tunnel engineering, 2021) and the code for design of Zhejiang province
FEM, while the openings of joints are slightly smaller than those of the urban rail transit (DB33/T 1146-2018 Code for design of Zhejiang
FEM. For soil with low stiffness, the Winkler-Timoshenko solutions province urban rail transit, 2018), the maximum opening and disloca­
tion reach up to 65% and 37% of the limit values, respectively. In terms

0.8 0.30
FEM FEM
Winkler solution Winkler solution
0.4
Dislocation/mm

0.15
Opening/mm

0.0

0.00
-0.4

-0.8 -0.15
-50 -25 0 25 50 -50 -25 0 25 50
x/m x/m
(a) Dislocation (b) Opening

Fig. 13. Calculated dislocation and opening of existing tunnel versus numerical result.

10
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

Opening of joint Dislocation between Opening of joint


at the crown rings at the crown with at the crown
no opening

(mm)

Dislocation between Opening of joint Dislocation between


rings at the invert at the invert rings at the invert

Es = 10 MPa Es = 20.2 MPa

Es = 65 MPa Es = 100 MPa


Fig. 14. Deformation contour of tunnel (scale up 500 times).

x/m 0.9 Style Color


-50 -25 0 25 50 FEM Es=10 MPa
-1 Winkler solution Es=20.2 MPa
0.6 Continuum solution Es=65 MPa
0 Es=100 MPa
Dislocation/mm

0.3
Displacement/mm

2 0.0

3
Style Color -0.3
4 FEM Es=10 MPa
Winkler solution Es=20.2 MPa -0.6
5 Continuum solution Es=65 MPa -50 -25 0 25 50
Es=100 MPa
6 x/m

Fig. 15. Comparison of tunnel displacements. Fig. 16. Comparison of dislocation between segment rings.

of the codes, it is necessary to strengthen the monitoring of the structure 4. Estimating the maximum opening and dislocation
and even take further protective measures. Note that the volume loss of
0.625% in this case induced a much larger deformation than those of the 4.1. Design charts for maximum bending moments and shear forces
previous case with a volume loss of 0.37%, which is mainly because the
diameter of this newly-built tunnel is two times the previous one. As previously mentioned, openings Δ and dislocations δ between
From the above discussion, the applicability of the equivalent stiff­ rings raised by the differential settlement have potential risks to the
nesses in the prediction of the opening and dislocation is finally verified tunnel’s structure due to tunneling underneath. In terms of Eqs. (25) and
by comparisons with a well-documented case study and the FEM for the (26), they can be converted to solve the maximum bending moment
problem of tunnel responses to tunneling underneath. Mmax and the maximum shear force Vmax of the continuous Timoshenko
beam model. In the previous sections, the Winkler-Timoshenko solution
and the equivalent longitudinal stiffness have been verified. This section
will further utilize the Winkler-Timoshenko solution to establish design
charts to predict the maximum bending moment Mmax and the maximum

11
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

0.4 Substituting the standard Gaussian curve, it is observed that the


Style Color
FEM Es=10 MPa maximum sagging bending moment Mmax tends to M*max ( = EISi2max ). The
0.3 Winkler solution Es=20.2 MPa normalized maximum bending moment thus has the form of M
M*
max
, termed
max
Continuum solution Es=65 MPa as the modification factor MF.
0.2 Es=100 MPa Following Klar’s (2005) normalization method, a series of para­
Opening/mm

metric studies are carried out based on the Winkler-Timoshenko solu­


0.1 tion. Referring to Table 1 and Table 2, three common tunnels with
different diameters of 6 m, 9 m, and 12 m are subjected to the Greenfield
0.0 conditions with varying values of i (3 m, 6 m, 12 m, 18 m and 24 m) and
α (0.01, 0.1, 0.3 and 0.5). A consistent Smax of 0.02m is used, as it would
-0.1 not affect the later normalized results. Each of them is placed with three
embedment depths of 10 m, 20 m, and 30 m. The shear factor Q varies
-0.2 from 0.1 to 100 and the elastic modulus of soil varies from 0.1 MPa to
-50 -25 0 25 50 200 MPa. A total of 5760 arrangements are included.
sag
Mmax
x/m Fig. 23(a) plots the data pairs of MFM,sag ( = sag,* ) versus R with the
Mmax
standard Greenfield condition (α = 0.5). It should be noted that the
Fig. 17. Comparison of the opening of joints.

shear force Vmax to rapidly assess the openings Δ and dislocations δ.


Based on the Euler-beam-based elastic continuum solution, Klar et al.
(2005) demonstrated the normalized maximum sagging bending mo­
2
ments M max i EI
EISmax and the relative bending stiffness R ( = Es r0 i3 ) have a unique
relationship when the greenfield has the maximum settlement Smax and
the distance of inflection point i, such that the maximum bending
moment Mmax can be easily determined in terms of the beam-soil relative
stiffness and greenfield condition.
It should be noted that the relationship established by Klar et al.
(2005) was based on the bending properties that, for an Euler beam
buried in very stiff soil, the bending moment values tend to those ob­
2
tained by forcing the beam to follow the soil (i.e. Mmax = − EI ∂ ∂wxCAT
2 ).
Fig. 19. Layout of existing tunnel and new tunnel.

1.0 1.0
Style Color Style Color
i/r0=1.94 Q=0.26 i/r0=1.94 Q=0.26
0.8 i/r0=2.15 Q=2.6 0.8 i/r0=2.15 Q=2.6
Shanghai M8 Shenzhen M3 Shanghai M8 Shenzhen M3
maxi /EISmax
i /EISmax

0.6 Es=10 MPa 0.6 Es=10 MPa


Es=20.2 MPa Es=20.2 MPa
2
sag 2

0.4 Es=65 MPa 0.4 Es=65 MPa


Mmax

M hog

Es=100 MPa Es=100 MPa


0.2 0.2

0.0 0.0
10-1 100 101 102 103 104 105 10-1 100 101 102 103 104 105
EI/Esr04 EI/Esr04

(a) Sagging bending moment (b) Hogging bending moment


1.0
Style Color
i/r0=1.94 Q=0.26
0.8 i/r0=2.15 Q=2.6
Shanghai M8 Shenzhen M3
Vmaxi/GAsSmax

0.6 Es=10 MPa


Es=20.2 MPa
0.4 Es=65 MPa
Es=100 MPa
0.2

0.0
10-1 100 101 102 103 104 105
EI/Esr04

(c) Shearing force

Fig. 18. The description of deformation mode.

12
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

Table 4 Fig. 23(d), all data points fall into a narrower band, when the modified
Material properties and geometric parameters for the tunnel structure of ( )0.3
relative bending stiffness has a form of RT 2Nα , which is further
Shanghai Metro Line 8. M,sag M,sag
defined as RTα . When α goes to 0.5, RTα returns to RT . The results
Lining ring Ec 3.45 × 104 MPa
from Franza and Viggiani (2021) are also provided as green squares in
v 0.2
D 6.2 m Fig. 23(d). It can be seen that these data points almost all fall within the
t 350 mm narrow band, illustrating the rationality of the normalization method.
Bolt Eb 2.06 × 105 MPa Regression analysis on the 5760 data points in the figure is then
v 0.2 carried out, and showed that they can be represented reasonably well by
ls 1.2 m
lb 0.4 m
the following equation:
Db 30 mm 1
nb 17
M,sag
MFav = ( )0.77 (27)
0.34 RM,sag
Tα +1

sag,* Fig. 23(d) shows this average bound curve as a solid curve. The
maximum sagging bending moments Mmax here are directly obtained by
equations for the upper and lower bound curves can be fitted as follows:
forcing the Timoshenko beam to follow the greenfield rather than using
EISmax
, as the expression is deduced from the Euler beam, without 1
i2 MFUM,sag = ( )0.77 (28)
considering the effect of the shear stiffness. It is evident from Fig. 23(a) 0.23 RM,sag
Tα +1
that the data points are scattered over a wide area.
A similar problem was also noticed by Franza et al. (2020) and Yu 1
MFLM,sag = ( )0.77 (29)
et al. (2022) when evaluating the modification factor for the building 0.5 RM,sag +1

façade in terms of the relative bending stiffness. Thus, Franza et al.
(2020) proposed a modified bending stiffness by matching the central Similarly, for the convenience of evaluating the maximum hogging
deflections of a simply supported Euler beam and a simply supported bending moment and maximum shearing force, Fig. 24(a) and (b) plot
hog ( )2.5
Timoshenko beam, such that the contribution to the deflection arising Mmax
the data pairs of MFM,hog (= hog,* ) versus RM,hog
Tα [=RT 2Nα ] and MFV (=VVmax
* )
Mmax
from shear flexibility has been accounted for. The expression of the
max
( )0.5
modified bending stiffness is presented as EI versus RVTα [=RT 2Nα ], respectively. Mhog,* max and Vmax are the maximum
*
2 , where a is a
1+a(i/Lsag/hog ) Q hogging bending moment and maximum shearing force obtained by
coefficient depending on the distribution of the soil-structure contact forcing the Timoshenko beam to follow the greenfield. It can be seen
force and the boundary condition; Lsag/hog is the transverse length of the that, compared with RM,sag and RVTα , RM,hog have a larger exponent for 2Nα,
Tα Tα
sagging/hogging region of the tunnel settlement trough. Utilizing the which implies the greenfield shape has a significant effect on the
modified bending stiffness, the Euler-beam-based relative bending
stiffness R is rewritten as the modified relative bending stiffness RT =
EI
[ 2
] to account for the effect of shear flexibility. x/m
1+a(i/Lsag/hog ) Q Es r0 i3
( )2 -50 -25 0 25 50
The finding from above is that a and i/Lsag/hog would change with
the actual input conditions. For convenience, the dimensionless group 0
( )2
factor a i/Lsag/hog is approximately determined to be 1 through trial
Displacement/mm

and error, as shown in Fig. 23(b), which gives rise to the smallest en­
10
velope for the data band when plotting for MFM,sag against RT . RT is then
EI
simplified as (1+Q)E s r0 i
3 . When Q approaches 0, RT returns to R.
20
Fig. 23(c) further investigates the effect of the shape of greenfield
settlement troughs. As mentioned earlier, for an Euler beam buried in
very stiff soil, the bending moment values tend to those obtained by 30 FEM
forcing the beam to follow the soil (i.e. Mmax = −
2
EI ∂ ∂wxCAT Winkler solution
2 ). Substituting
2α EISmax
Measurement
the modified Gaussian curve, M*max becomes N i2
. So, to better ac­ 40
count for the effect of the greenfield shape, the factor 2Nα is further
Fig.21. Calculated tunnel settlements versus measurement and numeri­
introduced for the modified relative bending stiffness RT . As shown in cal result.

Fig. 20. Deformation contour of tunnel (scale up 400 times).

13
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

3 2
FEM FEM
2 Winkler solution Winkler solution

Dislocation/mm
1 1

Opening/mm
0

-1 0

-2

-3 -1
-50 -25 0 25 50 -50 -25 0 25 50
x/m x/m

(a) Dislocation (b) Opening

Fig. 22. Calculated dislocation and opening of existing tunnel versus numerical result.

Fig. 23. Dimensionless plot of maximum sagging bending moment.

maximum hogging bending moment.


1
The equations of the average, upper and lower bound curves for MFLM,hog = ( )0.9 (32)
MFM,hog can be fitted as follows: 0.43 RM,hog
Tα +1

1 The equations of the average, upper and lower bound curves for MFV
M,hog
MFav = ( )0.9 (30) can be fitted as follows:
0.24 RM,hog
Tα +1
1
1
V
MFav = ( )0.86 (33)
MFUM,hog = ( ) (31) 0.55 RVTα +1
M,hog 0.9
0.17 RT α +1

14
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

Fig. 24. Dimensionless plot of maximum hogging bending moment and shearing force.

max , Mmax , Vmax and wbmax of Figs. 23-25 are obtained


Note that Msag,* hog,* * *

directly by the Winkler-Timoshenko solutions. For the practical appli­


cation of the design charts, it is necessary to provide simple formulas for
them. For a Euler beam with a standard Gaussian curve input, M*max is
deduced as EISi2max . Therefore, to consider the shearing effect and the
modified Gaussian curve, Msag,*
max , Mmax , Vmax and wbmax can be modified
hog,* * *

from the corresponding Euler-beam-based solutions using the factors 2Nα


and Q. They are approximately fitted as Eqs.(40)-(43). Fig. 26 shows the
effectiveness of the expressions.

EISmax 0.9(2α/N)1.4
sag,*
Mmax = 0.67 0.13
(40)
i2
(1 + Q) + (2α/N)Q(1 + Q)−

EISmax 0.35(2α/N)2.2
hog,*
Mmax = (41)
i2 (1 + Q) + (2α/N)Q(1 + Q)0.22
0.67

EISmax 1.2(2α/N)1.6
Fig. 25. Dimensionless plot of maximum bending displacement. *
Vmax = (42)
i3 1 + Q + (2α/N)Q

1
MFUV = ( )0.86 (34) w*bmax = Smax
0.92
(43)
0.4 RVTα +1 1 + (2α/N)Q(1 + Q)− 0.6

1 4.2. Application of the design charts


MFLV = ( )0.86 (35)
0.8 RVTα +1
This subsection will demonstrate the evaluation process of the
According to Eqs. (6b) and (8), the maximum total displacement wmax
maximum openings Δ and dislocations δ between rings based on the
can be expressed as follows:
design charts for maximum bending moments and shear forces through
the cases introduced in Section 3.4. The detailed evaluation procedure is
sag
Mmax
wmax = wbmax + (36)
κGA summarized as follows.
Msag
max can be calculated using Eqs. (27)-(29). As long as wbmax is
determined, wmax can be then evaluated. (1) Determining the basic parameters of the shield tunnel and soil:
( )2.5 tunnel outer diameter D, lining thickness t, lining width ls ,
Fig. 25 plots the data pairs of MFwb (=wwbmax ) and RwTαb [=RT 2Nα ].
*
bmax number of bolts nb , diameter of bolt Db , length of bolt lb , elastic
w*bmax is the maximum bending displacement obtained by forcing the modulus of the shield segment Ec , elastic modulus of the bolt Eb ,
Timoshenko beam to follow the greenfield, and the equations of the shear modulus of the tunnel segment Gc , shear modulus of the
average, upper and lower bound curves for MFwb can be expressed as bolt Gb , elastic modulus of soil Es and greenfield shape parame­
follows: ters (Smax , i and α).
(2) Determining the equivalent longitudinal bending stiffness EI and
1
wb
MFav = 0.48
(37) shear stiffness κGA by Eqs. (23) and (24).
0.11(RwT αb ) +1 hog,* sag,*
*
(3) Determining the Mmax , Mmax Vmax and w*bmax by Eqs. (40)-(43).
EI
1 Determining relative structure-soil stiffness RT = (1+Q)E s r0 i
3.
MFUwb = (38) ( 2α)0.3 ( 2α)2.5
0.1(RwT αb )0.48 + 1 M,sag M,hog
Determining RTα = RT N , RTα = RT N , RVTα =
( )0.5 ( )2.5
1 RT 2Nα and RwTαb = RT 2Nα . Substituting them into Eqs. (27)-
MFLwb = (39) hog
0.15(RwT αb )0.48 + 1 (39) to calculate the maximum sagging bending moment Mmax ,

15
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

Fig. 26. Fitting of Winkler-Timoshenko solution.

sag
the maximum hogging bending moment Mmax , the maximum proposed elastic continuum solution is validated by using the results of
shearing force Vmax and the maximum displacement wbmax . Franza and Viggiani (2021). The applicability of Yu et al’s (2013)
(4) Calculating the maximum openings Δ and dislocations δ between Winkler modulus in the longitudinal response of existing tunnels
rings and the maximum displacement by substituting Mmax , Vmax induced by tunneling underneath is verified against the elastic contin­
and wbmax into Eqs. (25), (26) and (36). uum solution. The Winkler-Timoshenko solution with Yu et al’s (2013)
Winkler modulus can provide similar maximum values as well as similar
Take the case study reported by Fan (2008) in Section 3.4 as an profiles to the continuum solution.
example to illustrate the applicability of design charts. Also, larger The rationality of Cheng et al.’s (2020) equivalent bending stiffness
volume loss (1% and 1.5%) is set to discuss its effects on the tunnel and Wu et al.’s (2015) equivalent shear stiffness is further demonstrated
deformation, especially the deformation of joints. Fig. 27 shows the through two field cases and elaborate finite element shield tunnel
comparison between the numerical and predicted maximum opening, models that consider the soil as the PSI element with the proposed
dislocation of rings and displacement of the tunnel. Section 3.3 illus­ Winkler spring stiffness. The verifications with the three-dimensional
trates that the Winkler-Timoshenko solutions provide an overestimated models demonstrate the applicability of the equivalent Timoshenko
prediction for settlement and an underestimated prediction for the beam for shield tunneling.
maximum dislocation and opening between rings, such that the average The framework of Franza and Viggiani’s (2021) for longitudinal
prediction of the design charts established from the Winkler- tunneling is validated, which also demonstrates that the shear factor Q
Timoshenko solutions overestimates the deformation of the tunnel and can be used to govern the relative importance of shear and bending
underestimates the deformation of the joints. Furthermore, the com­ deformations and account for the settlement trough width. Based on
parison with the limiting values shows that, for the present case, it is their work, the modified relative bending stiffness RTα accounting for the
necessary to control the volume loss within 1% to ensure no excessive effect of the shear stiffness and modified Gaussian curves is proposed.
deformation occurs at the joints. Using the Winkler-Timoshenko solution, four design charts are estab­
lished for the maximum sagging bending moment, the maximum hog­
5. Conclusions ging bending moment, the maximum shearing force and the maximum
displacement, respectively. Comparisons with the field case and nu­
This study establishes the elastic continuum and Winkler solutions merical results reveal that the present Winkler-Timoshenko solution
based on the displacement-based single variable Timoshenko-beam would predict a slightly larger maximum settlement and slightly smaller
governing differential equation that can solve the shear locking. The maximum opening of the joints and maximum dislocation between
rings. Therefore, the upper-bound curves in the design charts are

16
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

Fig. 27. Comparison between the maximum numerical and predicted results.

recommended to estimate the maximum opening and dislocation and Acknowledgments


the lower-bound curve is recommended to estimate the maximum tun­
nel displacement. The authors acknowledge the financial support from the National
In addition, Franza and Viggiani (2021) also discussed the tunnel Natural Science Foundation of China (Grant 51738010) and the Na­
responses due to twin tunneling by superimposing two standard tional Key R&D Program of China (Grant 2016YFC0800200).
Gaussian curves. This can be a basis of extending the present work to
further consider the effect of twin tunneling with different spacings and References
volume losses.
Attewell, P.B., Yeates, J., Selby, A.R., 1986. soil movements induced by tunnelling and
their effects on pipelines and structures. Blackie and Son Ltd., London.
CRediT authorship contribution statement Basmaji, B., Deck, O., Al Heib, M., 2019. Analytical model to predict building deflections
induced by ground movements. Eur. J. Environ. Civ. Eng. 23 (3), 409–431.
Jian Yu: Methodology, Software, Investigation, Writing – review & Chen, R.-P., Chen, S., Wu, H.-N., Liu, Y., Meng, F.-Y., 2020. Investigation on deformation
behavior and failure mechanism of a segmental ring in shield tunnels based on
editing. Hao Li: Methodology, Software, Writing – original draft, Visu­ elaborate numerical simulation. Eng. Fail. Anal. 117.
alization. Maosong Huang: Conceptualization, Methodology, Writing – Chen, S., Wu, H.N., Chen, R.P., Shen, S.L., Liu, Y., 2021. Deformation of collinear tunnel
review & editing, Supervision, Funding acquisition. Yonghui Li: induced by overlying long-distance excavation. J. Shanghai Jiaotong Univ. 55 (6),
699–706 in Chinese.
Conceptualization, Supervision. Jorgin Qi Wen Tan: Resources, Cheng, H., Chen, R., Wu, H., Meng, F., Yi, Y., 2021. General solutions for the longitudinal
Writing – review & editing. Yuancheng Guo: Conceptualization, deformation of shield tunnels with multiple discontinuities in strata. Tunn. Undergr.
Supervision. Sp. Tech. 107.
Franza, A., Acikgoz, S., DeJong, M.J., 2020. Timoshenko beam models for the coupled
analysis of building response to tunnelling. Tunn. Undergr. Sp. Tech. 96.
Declaration of Competing Interest Fan, Y.Y., 2008. Study on the small disturbance control for shield crossing adjacent
running subway. Thesis, Tongji University (in Chinese), M. Sc.
The authors declare that they have no known competing financial Franza, A., Viggiani, G.M.B., 2021. Role of shear deformability on the response of tunnels
and pipelines to single and twin tunneling. J. Geotechn. Geoenviron. Eng. ASCE. 147
interests or personal relationships that could have appeared to influence (12), 04021145.
the work reported in this paper. Goh, K.H., Mair, R.J., 2014. Response of framed buildings to excavation-induced
movements. Soils Found. 54 (3), 250–268.
Gue, C.Y., Wilcock, M., Alhaddad, M.M., Elshafie, M.Z.E.B., Soga, K., Mair, R.J., 2015.
The monitoring of an existing cast iron tunnel with distributed fibre optic sensing
(DFOS). J. Civ. Struct. Health Monit. 5 (5), 573–586.

17
J. Yu et al. Computers and Geotechnics 147 (2022) 104757

Gu, X.Q., Wu, R.T., Liang, F.Y., Gao, G.Y., 2021. On HSS model parameters for Shanghai Mair, R.J., Taylor, R.N., Bracegirdle, A., 1993. Subsurface settlement profiles above
soils with engineering verification. Rock Soil Mech. 42 (3), 833–845. In Chinese. tunnels in clays. Geotechnique 43 (2), 315–320.
Huang, M.S., Zhang, C.R., Li, Z., 2009. A simplified analysis method for the influence of Ma, S.K., Shao, Y., Liu, Y., Jiang, J., Fan, X.L., 2017. Responses of pipeline to side-by-side
tunneling on grouped piles. Tunn. Undergr. Sp. Tech. 24 (4), 410–422. twin tunnelling at different depths: 3D centrifuge tests and numerical modelling.
Huang, M.S., Zhou, X.C., Yu, J., Leung, C.F., Tan, J.Q.W., 2019. Estimating the effects of Tunn. Undergr. Sp. Tech. 66, 157–173.
tunnelling on existing jointed pipelines based on Winkler model. Tunn. Undergr. Sp. GB/T 51438-2021 Standard for design of shield tunnel engineering. Beijing: China
Tech. 86, 89–99. Architecture and Building Press; 2021. (in Chinese).
Huang, F.u., Zhang, M., Wang, F., Ling, T., Yang, X., 2020. The failure mechanism of Peck, R.B., 1969. Deep excavations and tunnelling in soft ground. In: In: Proceedings of
surrounding rock around an existing shield tunnel induced by an adjacent 7th International Conference of Soil Mechanics and Foundation Engineering, State of
excavation. Comput. Geotech. 117. the Art, pp. 225–290.
Jia, R.H., Yang, J.S., Ma, T., Liu, S.Y., 2009. Field monitoring and numerical analysis of Shiba, Y., Kawashima, K., Obinata, N., Kano, T., 1988. An evaluation method of
shield tunneling considering existing tunnels. Chin. J. Geotech. Eng. 31(3), 425-430 longitudinal stiffness of shield tunnel linings for application to seismic response
(In Chinese). analyses. Doboku Gakkai Ronbunshu 1988 (398), 319–327. In Japanese.
Jin, D.L., Yuan, D.J., Li, X.G., Zheng, H.T., 2018a. An in-tunnel grouting protection Shen, S.L., Wu, H.N., Cui, Y.J., Yin, Z.Y., 2014. Long-term settlement behaviour of metro
method for excavating twin tunnels beneath an existing tunnel. Tunn. Undergr. Sp. tunnels in the soft deposits of Shanghai. Tunn. Undergr. Sp. Tech. 40, 309–323.
Tech. 71, 27–35. Shi, J.W., Wang, Y., Ng, C.W.W., 2016a. Three-dimensional centrifuge modeling of
Jin, D.L., Yuan, D.J., Li, X.G., Zheng, H.T., 2018b. Analysis of the settlement of an ground and pipeline response to tunnel excavation. J. Geotechn. Geoenviron. Eng.
existing tunnel induced by shield tunneling underneath. Tunn. Undergr. Sp. Tech. ASCE. 142 (11).
81, 209–220. Shi, C.H., Cao, C.Y., Lei, M.F., Peng, L.M., Ai, H.J., 2016b. Effects of lateral unloading on
Klar, A., Vorster, T.E.B., Soga, K., Mair, R.J., 2005. Soil-pipe interaction due to the mechanical and deformation performance of shield tunnel segment joints. Tunn.
tunnelling: comparison between Winkler and elastic continuum solutions. Undergr. Sp. Tech. 51, 175–188.
Geotechnique 55 (6), 461–466. Timoshenko, S.P., 1921. On the correction for shear of the differential equation for
Kiendl, J., Auricchio, F., Hughes, T.J., Reali, A., 2015. Single-variable formulations and transverse vibration of prismatic bars. Philo. Mag 41, 744.
isogeometric discretizations for shear deformable beams. Comput. Methods Appl. Tomaževič, M., 2009. Shear resistance of masonry walls and Eurocode 6: shear versus
Mech. Engrg. 284, 988–1004. tensile strength of masonry. Mater. Struct. 42 (7), 889–907.
Liao, S.M., Peng, F.L., Shen, S.L., 2008. Analysis of shearing effect on tunnel induced by Vorster, T.E., Klar, A., Soga, K., Mair, R.J., 2005. Estimating the effects of tunneling on
load transfer along longitudinal direction. Tunn. Undergr. Sp. Tech. 23 (4), 421–430. existing pipelines. J. Geotechn. Geoenviron. Eng. ASCE. 131 (11), 1399–1410.
Li, P., Du, S.J., Shen, S.L., Wang, Y.H., Zhao, H.H., 2016. Timoshenko beam solution for Vesić, A.B., 1961. Bending of beams resting on isotropic elastic solid. J. Soil Mech.
the response of existing tunnels because of tunneling underneath. Int. J. Numer. Found. Eng., ASCE. 87 (2), 35–53.
Anal. Meth. Geomech. 40 (5), 766–784. Wang, Z.X., Miao, L.C., Wang, R.R., Wang, F., Wang, X.L., 2014a. Physical model tests
Liang, R.Z., Xia, T.D., Huang, M.S., Lin, C.G., 2017. Simplified analytical method for and PFC3D modeling of soil-pipe interaction in sands during tunnelling. Chin. J.
evaluating the effects of adjacent excavation on shield tunnel considering the Geotech. Eng. 36 (1), 182–188 in Chinese.
shearing effect. Comput. Geotech. 81, 167–187. Wang, Z., Wang, L.Z., Li, L.L., Wang, J.C., 2014b. Failure mechanism of tunnel lining
Li, M.G., Chen, J.J., Wang, J.H., Zhu, Y.F., 2018. Comparative study of construction joints and bolts with uneven longitudinal ground settlement. Tunn. Undergr. Sp.
methods for deep excavations above shield tunnels. Tunn. Undergr. Sp. Tech. 71, Tech. 40, 300–308.
329–339. Wu, H.N., Shen, S.L., Liao, S.M., Yin, Z.Y., 2015. Longitudinal structural modelling of
Liang, R.Z., 2019. Simplified analytical method for evaluating the effects of overcrossing shield tunnels considering shearing dislocation between segmental rings. Tunn.
tunnelling on existing shield tunnels using the nonlinear Pasternak foundation Undergr. Sp. Tech. 50, 317–323.
model. Soils Found. 59 (6), 1711–1727. Wham, B.P., Argyrou, C., O’Rourke, T.D., 2016. Jointed pipeline response to tunneling-
Lin, X.T., Chen, R.P., Wu, H.N., Cheng, H.Z., 2019. Deformation behaviors of existing induced ground deformation. Can. Geotech. J. 53 (11), 1794–1806.
tunnels caused by shield tunneling undercrossing with oblique angle. Tunn. Undergr. Xu, L., 2005. Study on the longitudinal settlement of shield tunnel in soft soil. Tongji
Sp. Tech. 89, 78–90. University (in Chinese). Ph.D. Thesis.
Lin, C., Huang, M., Nadim, F., Liu, Z., 2020. Embankment responses to shield tunnelling Yu, J., Zhang, C.R., Huang, M.S., 2013. Soil-pipe interaction due to tunnelling:
considering soil-structure interaction: case studies in Hangzhou soft ground. Tunn. assessment of Winkler modulus for underground pipelines. Comput. Geotech. 50,
Undergr. Sp. Tech. 96. 17–28.
Liu, D., Wang, F., Hu, Q., Huang, H., Zuo, J., Tian, C., Zhang, D., 2020. Structural Yu, J., Leung, C.F., Huang, M., Tan, J.Q.W., 2022. Assessment of settlement-based strain
responses and treatments of shield tunnel due to leakage: a case study. Tunn. in masonry building facade due to tunneling. Comput. Geotech. 144.
Undergr. Sp. Tech. 103. Zhang, C.R., Yu, J., Huang, M.S., 2012. Effects of tunnelling on existing pipelines in
Liu, D., Tian, C., Wang, F., Hu, Q., Zuo, J., 2021. Longitudinal structural deformation layered soils. Comput. Geotech. 43, 12–25.
mechanism of shield tunnel linings considering shearing dislocation of Zhang, Z.G., Huang, M.S., 2014. Geotechnical influence on existing subway tunnels
circumferential joints. Comput. Geotech. 139. induced by multiline tunneling in Shanghai soft soil. Comput. Geotech. 56, 121–132.
Mindlin, R.D., 1936. Force at a point in the interior of a semi-infinite solid. Physics. 7 (5), DB33/T 1146-2018 Code for design of Zhejiang province urban rail transit. Beijing:
195–202. China Communications Press; 2018. (in Chinese).

18

You might also like