You are on page 1of 11

International Journal of Rock Mechanics & Mining Sciences 77 (2015) 328–338

Contents lists available at ScienceDirect

International Journal of
Rock Mechanics & Mining Sciences
journal homepage: www.elsevier.com/locate/ijrmms

Normal and shear resistance of longitudinal contact surfaces


of segmental tunnel linings
Akbar Salemi a,n, Morteza Esmaeili b, Farhang Sereshki a
a
Department of Mining, Petroleum and Geophysics, Shahrood University of Technology, University Boulevard, P.O. BOX 316, Shahrood, Iran
b
School of Railway Engineering, Iran University of Science and Technology, Narmak Tehran 16846, Iran

art ic l e i nf o a b s t r a c t

Article history: Stiffness of contact points has a significant effect on the internal forces and structural design of segmental
Received 24 July 2014 lining of mechanically bored tunnels. Recent research on the effect of contact points in the structural
Received in revised form analysis of segmental lining can be categorized into: (1) using hinges at contact points (HL model),
10 November 2014
(2) reducing liner rigidity (RR model), (3) using effective moment of inertia for liners (EMI model), (4) using
Accepted 27 April 2015
rotational springs at contact points (BRS model), and (5) developing a comprehensive 3D model for
Available online 17 May 2015
segmenting contact locations. The present study tried to develop a precise contact model based on
Keywords: experimental direct shear tests (DST) on the concrete samples of contact points in segments. By selecting
Mechanized tunneling the normal stress between 0.25 and 2 MPa, about 90 tests were performed on the grooved cubic samples
Segmental lining
with and without gasket. As the practical outcome of this study, the contact shear and normal reaction
Longitudinal joints
moduli ks and kn were related to contact normal stress via two linear regression equations considering R2 of
Contact stiffness
Direct shear test 99%. For evaluating the proposed method, finite element models of an urban tunnel liner were developed
using the concept of beam on elastic foundation considering the proposed contact model of the present
study (beam-contact springs (BCS)), the results of which were compared with those of the conventional
contact models. Results demonstrated that the proposed model of this research had the highest
correspondence with reduced RR, HL, EMI, BRS, and comprehensive coverage, respectively.
& 2015 Elsevier Ltd. All rights reserved.

1. Introduction In this paper mechanical behavior of longitudinal joints in


segmental lining was studied in order to find a definite relationship
Segmental lining has been extensively applied in mechanized between stiffness of contact points and normal stress in contact
shield tunneling in recent years, and the shield tunneling method locations. For this purpose, direct shear test (DST) apparatus was
has been widely adopted for designing and constructing urban utilized for testing real concrete samples in order to obtain normal
underground tunnels in soft grounds due to its flexibility, cost and shear stiffness coefficients of longitudinal contact surface of the
effectiveness, minimum impact on urban traffic, and settlement segmental lining. According to the standard of International Society
risk on surface structures. A large part of the cost of shield of Rock Mechanics [1], the tests were performed on the concrete
tunneling construction is related to segment manufacturing; samples of the segment contact surface in two cases of with and
hence, efficient design of these elements can considerably affect
without gasket. In this manner, 46 concrete samples were prepared
total construction cost. In this regard, understanding the actual
from both A and B segments (two consecutive segments with a
behavior of the segmented liner under surrounding loads is an
shared contact surface in the same ring, as shown in Fig. 1). The
important issue from both structural analysis and design stand-
samples were tested under different normal stresses (σ n ) of 0.25–
points. In other words, as the lining of a shield tunnel cannot be
2 MPa and variation of normal and shear stresses (Δσ n , Δτ) were
assumed as a continuous ring due to the existence of longitudinal
joints, effect of segment-to-segment contact should be properly recorded based on normal and shear displacements (Δδn , Δδs ).
considered during its structural analysis in order to access the Consequently, the values of kn ¼ Δσ n =Δδn and ks ¼ Δτ=Δδs as normal
realistic values of internal forces and displacements. and shear reaction moduli were presented as a function of σ n . At the
final stage, the derived equations for the normal and shear reaction
moduli were converted into normal and shear stiffness and were
implemented at the contact point of a numerical example as a case
n
Corresponding author. Tel.: þ 98 21 66836033; fax: þ 98 21 77491201. study of an urban tunnel. Results of numerical models including the
E-mail address: akbar.salemi@gmail.com (A. Salemi). liner internal forces and deformations were obtained using the

http://dx.doi.org/10.1016/j.ijrmms.2015.04.014
1365-1609/& 2015 Elsevier Ltd. All rights reserved.
A. Salemi et al. / International Journal of Rock Mechanics & Mining Sciences 77 (2015) 328–338 329

Fig. 1. Geometry and arrangement of the segments in a ring.

contact model developed in the present study, called beam-contact of inertia for the liners of jointed shallow tunnels as follows [16]:
springs (BCS), and compared with those obtained by implementing
I e; max =I ¼ 429:01n  4:6023 ð2Þ
other methods such as united lining (UL), reduced rigidity of lining
(RR), hinged lining (HL), effective moment of inertia (EMI), and
I e; min =I ¼ 159:19n  4:2734 ð3Þ
beam-rotational spring (BRS).
where I e;max and I e;min are effective moment of inertia of lining with
respect to critical and most favorable orientation of joints, respec-
tively, and I is moment of inertia for non-jointed lining.
2. Literature survey
Using beam rotational spring model (BRS model): In this model,
segments are modeled by beam elements and longitudinal joints
Reviewing the available technical literature on this issue
using rotational springs. Many researchers have proposed various
reveals four main categories of contact models for assessing the
values for rotational stiffness, kθ . For instance, Mashimo and
effects of longitudinal joints on liner internal forces as follows:
Ishimura suggested the range of 32–127 MN m/rad for this para-
Using hinges at liner contact points (HL model): In this method,
meter [18]. Lee et al. proposed the range of 4–30 MN m/rad for such
segments are modeled using beam elements considering perfect
stiffness, which was based on the field measured values reported by
hinges in their contact locations. Assuming perfect joints in the
Chen and Zhou in Shanghai subway tunnel [19]. Koyama proposed
liner causes a pure axial force in the joint location without any
three diagrams for bending moment versus rotational angle (M θ)
bending moment. The joint behavior in this model cannot consider
behavior of contact joints in segmental lining [20]. In their study,
the flexural behavior of segment-to-segment contact area and
approximate values of rotational spring stiffness kθ as the slope of
leads to the transfer of bending moment to segments. This method
Mθ diagrams varied from 15 to 150 MN m/rad. Teachavorasinskun
usually causes more bending moment in the liner than other
and Chub-uppakarn, according to the experimental tests and FEM
contact models. Tang [2] and Zhong et al. [3] have utilized this
modeling, suggested the range of 1–3 MN m/rad for kθ [21]. In
method for structural analysis of segmental lining.
another study, Arnau and Molins presented the rotational stiffness of
Reducing liner rigidity (RR model): Some researchers and
segments as 50 and 100 MN m/rad with respect to two values of
societies such as Morgan [4], Peak et al. [5], Japan Society of Civil
compressive stress as 1.5 and 3 MPa in the contact location [22]. Do
Engineers (JSCE) [6], Ranken et al. [7], Einstein and Schwartz [8],
et al. simulated segment connections with perfect hinges, reduced
Yuen [9], Ogawa [10], Liu and Hou [11], Lee and Ge [12], Blom [13],
lining thickness in the narrowest part of the joint, and consequently
and El Naggar et al. [14] have considered segmental tunnel lining
calculated the bending moment of yielding condition (Myield) as
as a continuous ring with discounted rigidity by applying a
about 150 kN m/m [23]. This moment was in agreement with the
reduction factor, η r 1, to the bending stiffness (EI) of liners.
maximum permitted angular rotation of 0.01 rad (E1%) in the joint
Using effective moment of inertia of liners (EMI model): Muir
location. Based upon the reported values of Myield and the permitted
Wood proposed this method for the segmental lining of tunnels
rotation, rotational stiffness coefficient was calculated as about
considering the same length for each segment in the case of
100 MN m/rad/m based on 0:8ðM yield =θÞ. Overall, based on the
segment number of n 4 4 and introduced the effective moment of
aforementioned works, it seems that it is impossible to propose a
inertia of liners, Ie, as follows [15]:
unique value for rotational spring stiffness and this parameter can be
I e ¼ I j þ ð4=n2 ÞI ð1Þ selected in a wide range of 1–150 MN m/rad.
By summarizing the above-mentioned models, it could be
where I j is moment of inertia of liners in the joint location stated that the first method does not simulate properly the actual
(effective contact area) and I is moment of inertia of liners. construction conditions and considers joints as a hinge while
This method has been used by Lee and Ge [12], Hefny and Tan ignoring partial moment transmitting capacity. On the other hand,
[16] and Hefny and Chua [17] as a simple methodology for although effective bending rigidity must only affect contact points,
determining the stress induced in jointed lining without incorpor- in the second model, it affects the entire lining by ratio of the
ating the joints into structural analysis. In another work, Hefny bending rigidity [20]. In the third model, it is assumed that all
et al. proposed the maximum and minimum equivalent moment segments have the same length in a ring, which is not correct in
330 A. Salemi et al. / International Journal of Rock Mechanics & Mining Sciences 77 (2015) 328–338

most of practical cases. In the fourth method, rotational stiffness, longitudinal and circumferential joints [28]. They applied bitumen
kθ ; of tunnel lining joints is taken from a vast domain of values and rubber packers rather than concrete at contact points. Overall,
which may produce different results from those of others [19]. it is obvious that none of the aforementioned studies have
Stress and strain distribution over the contact point of segments is considered the effect of normal contact stress on normal and
controlled by a combination of shear, bending, and axial forces. In shear reaction modulus at of contact points of segmental liners.
sum, it can be stated that all of the aforementioned methods have
ignored the relevance of normal and tangential stiffness of contact
points to the normal stress acting on contact surfaces.
Some laboratory and numerical efforts have been made by
many researchers to model the real conditions of contact points. In 3. Laboratory testing program
this regard, Mayer et al. carried out some small-scale tests on
cam–pocket couplings and characterized shear and normal stiff- The main test program of the present study was founded on the
ness coefficients of two concrete segment joints with the bitumen utilized segment in water conveying tunnel in South of Tehran
rubber strips in between (concrete–bitumen rubber strips–con- called Abuzar tunnel. The tunnel was bored using an EPB (Earth
crete contact). They categorized the test results by normal contact Balanced Pressure) machine with the length of 4.2 km and
stress in the range of 0–0.25 MPa, 0.25–0.64 MPa, and more than diameter of 4.35 m. The tunnel permanent support was segmental
0.64 MPa and proposed normal reactions in these ranges of 0.34, lining with 1 þ5 arrangement, which was 0.25 m in thickness and
1.68, and 2.90 MPa/mm, respectively. Nevertheless, they presented 1.2 m in width. Inner diameter of the tunnel was 3.75 m and the
0.01 MPa/mm as shear stiffness for all of the mentioned ranges outer one was 4 m. This lining had 4 types of segments (A, B, C,
[24]. In 2009, Cavalaro studied the behavior of packers in segment- and K), as illustrated in Fig. 1. The contact area for all the segment
to-segment contact [25]. Later, Molins and Arnau, used Cavalaro's types was the same as seating dimensions 0.25  1.2 m2. The
constitutive model for fitting packer rings for normal compressive tunnel's overburden varied from 4 to 30 m. Table 1 summarizes
stress of 15 MPa to obtain variation of maximum shear stress characteristics of the tunnel lining along with its surrounding soil.
versus shear displacement; shear reaction was 10 MPa/mm in this In order to assess the mechanical behavior, many concrete
normal stress [26]. Do et al. [23] investigated factors affecting samples of segments' contact points were constructed in segment
segmental lining based on the work by Cavalaro and Aguado [27] mold and were tested in the direct shear test apparatus as shown
about the influence of packer on the contact points of segmental in Fig. 2. Due to double number of contact between A and B
lining of Barcelona L9 tunnel. In their study, in two conditions of segments in the segmental ring, this contact point was selected for
bitumen and rubber packer in filled and no packer under the sample manufacturing (Fig. 1). The produced samples from Type A
constant normal stress (2.75 MPa), the normal and shear stiffness and B segments had a contact surface with the dimensions of
were obtained respectively in the range of 100–3000 MN/m and 21 cm in length in thickness direction and 17 cm in width in
130–5500 MN/m during a series of laboratory tests. Do et al. longitudinal direction. As shown in Fig. 2, sample construction was
developed a 3D numerical model to use two values of 500 and performed using prismatic steel mold installed in interior surface
1050 MN/m for normal and shear stiffness, respectively, for both of segment molds at segment factory.

Table 1 Table 2
Characteristics of tunnel lining and the surrounding soil. Specifications of the concrete used for the produc-
tion of segment samples in contact location.
Soil Very silty clayey sand with gravel (passing-200¼ 30–60%)
Characteristic parameters Value
Natural density (kN/m3) 18.15
Elastic modulus (MPa) 44.13 Weight of Portland cement-type V (kg) 480
Poisson's ratio 0.32 Sand (%) 55
Friction angle (deg.) 32 Gravel (%) 45
Cohesion (kPa) 19.61 Water/cement ratio 0.32
Coefficient of earth pressure at rest 0.4 Super-plasticizer (%) 0.35
Slump–5 min (mm) 90
0
Lining Reinforced concrete (f c ¼ 51:09 MPa and f y ¼ 0:39 GPa) Slump–10 min (mm) 75
Elastic modulus (GPa) 29.62 1-day strength (MPa) 17.75
Poisson's ratio 0.2 11-day strength (MPa) 51.09
42-day strength (MPa) 71.39

Fig. 2. Details of the junction segment into another segment in a ring (a), sampling from inner surface of the segment mold (b), and detail of samples (c).
A. Salemi et al. / International Journal of Rock Mechanics & Mining Sciences 77 (2015) 328–338 331

3.1. Concrete characteristics

Specifications of the concrete used in the production of the


segment samples in contact location are summarized in Table 2.
The concrete samples were put for 42 days into the curing water
pool. Characteristic compressive strength of the concrete in the
segment production was 55.02 MPa and the mean target strength
was 56 MPa for the cubic samples with the size of 150 mm at the
age of 42 days. Concerning sulfate contaminant in the surrounding
soil of the tunnel, type V Portland cement was used in the concrete
mixture. In order to ensure the adequate strength and durability of
concrete, water/cement ratio was set to 0.32 and ratio of fine to
coarse aggregate was taken up 0.45–0.55 to ensure high perfor-
mance of polycarboxylate super-plasticizer as a concrete admix-
ture. Fresh concrete showed a slump between 5 and 8 cm.
Consequently, the one-day compressive strength of concrete was
obtained as 11.28 MPa, which was adequate for opening the mold
and transferring the samples to the curing water pool.

3.2. Gasket characteristics

In the construction process of concrete samples, waterproofing


of the joints between two segments was performed by a pair of
Fig. 3. Prepared concrete test samples with and without gasket for DSTs. 86-259 profile type of gaskets made of special rubber (EPDM;
Ethylene Propylene Diene Monomer (M-class) rubber) [29]. As
shown in Fig. 3, two produced concrete samples of A and B
segments were prepared for DST in two conditions; i.e. with or
without gasket.

3.3. Testing equipment

In DST setup, two shear boxes with the contact surface dimen-
sions of 17 cm in width and 21 cm in length are mostly used. The
applied testing equipment of this investigation had three cylind-
rical hydraulic jacks with 5.02 cm internal diameter and 50 kN
maximum load capacity. The horizontal and vertical gauges with
the accuracy of 0.01 mm were used to record the displacement of
samples. Fig. 4 illustrates details of the testing equipment.
The range of normal stress at the contact surface of the
segments could be easily calculated using thin membrane equa-
tion in plane stress conditions as follows:
σ n ¼ Pr=t ð4Þ
where P is hydrostatic pressure around tunnel due to overburden,
and r and t are inner radius of tunnel and thickness of lining,
respectively. Considering density of soil as 20 kN/m3, for 30 m
depth of overburden, the hydrostatic pressure P can be calculated
as 600 kN/m2 and consequently using Eq. (4), the normal stress on
the contact surface of segments σ n can be estimated as 4.89 MPa.
Fig. 4. Testing equipment parts of DST. In these experiments, due to the limitation of maximum imposed

4S 5S 7S 8S
5S - G 10S-G 12S-G 17S-G
9S 12S 14S 15S
1.0 1.0
Normal Stress (MPa)
Normal Stress (MPa)

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.0 0.1 0.2 0.3 0.4

Normal Displacement (mm) Normal Displacement (mm)


Fig. 5. Normal stress–normal displacement path with respect to the maximum normal stress of 1 MPa. (a) samples without gasket and (b) samples with gasket.
332 A. Salemi et al. / International Journal of Rock Mechanics & Mining Sciences 77 (2015) 328–338

4S 5S 7S 8S 5S-G 10S-G 12S-G 17S-G


9S 12S 14S 15S
0.8
0.8

Shear Stress (MPa)


Shear Stress (MPa)

0.6
0.6

0.4
0.4

0.2
0.2

0.0
0.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0

Shear Displacement (mm) Shear Displacement (mm)

Fig. 6. Graph of shear stress–shear displacement at the constant normal stress of 1 MPa. (a) samples without gasket and (b) samples with gasket.

Ks Kn Linear (Ks) Linear (Kn) Ks Kn Linear (Ks) Linear (Kn)

Reaction modulus (MPa/mm)


Reaction modulus (MPa/mm)

3.0
3.0

2.0
2.0

1.0
1.0

0.0 0.0
0.25 0.5 0.75 1 1.25 1.5 1.75 2 0.25 0.5 0.75 1 1.25 1.5 1.75 2

σn (MPa) σ n (MPa)

Fig. 7. Variation of shear and normal reaction modulus with normal stress on the contact surface. (a) samples without gasket and (b) samples with gasket.

load by DST jacks, vertical stress range was considered in the range Ks Kn
Difference between with and without Gasket

of 0.25–2 MPa. As will be shown in the next section, this limitation 20%
cannot affect the obtained linear equations for normal and shear
reaction coefficients with normal stress. 15%

10%
4. Results and discussion

5%
As mentioned before, two series of concrete samples were
prepared with/without gasket and examined under a variety of
normal stresses in the range of 0.25–2 MPa (by the increasing rate 0%
0.25 0.5 0.75 1 1.25 1.5 1.75 2
of 0.25 MPa). During DST on concrete samples A and B, first, the
σn (MPa)
vertical displacement gauge was set to zero and then the vertical
displacements were recorded parallel to applying the desired Fig. 8. Difference in the percentage of shear and normal reaction modulus in the
vertical load. In continue, maintaining the vertical load at a conditions of without and with gasket.

constant value, the horizontal load was gradually increased and


both horizontal and vertical displacements of the concrete sample and transfer it to the concrete part of segments, while its effect on
were recorded. shear stress distribution over the contact area can be under-
In the next section, the obtained test results for normal stress standable at low vertical stress. During the experiments, a small
σ n ¼ 1 MPa will be presented in the form of diagrams of normal number of samples was broken, the results of which were omitted
stress versus normal displacement and shear stress versus shear while drawing the relevant diagrams (samples 4S–15S and sam-
displacement for both samples with/without gasket. In this regard, ples 5S-G–17S-G). Results of all successful tests are presented in
a naming system was used for the samples; for instance, 5S refers Figs. 5 and 6 in the form of normal stress–normal displacement
to the fifth sample without gasket and 5S-G denotes the fifth and shear stress–shear displacement. As displayed in Fig. 5, varia-
sample with gasket. tion of normal displacement versus normal stress for both samples
without/with gasket were approximately linear and the average
4.1. Test results of both samples without/with gasket slope of the line represented the normal reaction coefficient, kn.
As shown in Fig. 6, shear stress versus shear displacement for
In DST on the samples without gasket, the effective area of both samples without/with gasket had a linear trend before full
contact surface was 17  15 cm2 during the compressive and shear sliding and the slope of this line expressed shear reaction
loading. As shown in Fig. 3, gasket profile 86-259 was glued using coefficient (ks ¼ ðΔτ=Δδs Þ). In the normal stress range of 0.25–
a special gum in the groove of the concrete samples. In this case, 2 MPa, shear stress threshold of sliding of the samples without
the effective contact area increased to 18  17 cm2. It should be gasket varied from 0.2 to 1.44 MPa while the sliding displacement
mentioned that, in practice, gasket cannot bear high normal stress changed from 0.45 to 0.58 mm and in order that, the threshold
A. Salemi et al. / International Journal of Rock Mechanics & Mining Sciences 77 (2015) 328–338 333

Fig. 9. Beam bedded on elastic foundation model with detail of joints.

-70 30
Axial Shear Bending 30
Force Force Moment
-100 15 15

-130 0 0

-160 -15 -15

-190 -30 -30

0.1 0.2 0.3 0.1 0.2 0.3 0.1 0.2 0.3


0.4 0.5 0.6 0.4 0.5 0.6 0.4 0.5 0.6
0.7 0.8 0.9 0.7 0.8 0.9 0.7 0.8 0.9

Horizontal 15.0 Vertical 15.0


Displacement 10.0 Displacement 10.0
5.0 5.0
0.0 0.0
-5.0 -5.0
-10.0 -10.0
-15.0 -15.0

0.1 0.2 0.3 0.1 0.2 0.3


0.4 0.5 0.6 0.4 0.5 0.6
0.7 0.8 0.9 0.7 0.8 0.9

Fig. 10. Axial and shear forces, bending moment, and vertical and horizontal displacements with different values of η in RR models.

shear stress with respect to sliding of the samples with gasket Based on the described procedure to obtain kn , and ks values for
varied from 0.15 to 1.42 MPa and its relevant shear displacement σ n ¼ 1 MPa, all of the extracted normal and shear reaction coeffi-
changes from 0.49 to 0.54 mm. cients for the whole range of σ n in both samples without/with
334 A. Salemi et al. / International Journal of Rock Mechanics & Mining Sciences 77 (2015) 328–338

gasket are illustrated in Fig. 7. In this figure, for the samples coefficients should be multiplied by the segments’ contact areas
without gasket, values of the shear and normal reaction coeffi- to be converted into normal and tangential springs.
cients varies from 0.25 to 2.79 and 0.96 to 2.85 MPa/mm and for
the samples with gasket, shear reaction coefficient varies from
0.21 to 2.78 MPa/mm and normal reaction coefficient changes 4.2. Comparing test results of the samples without and with gasket
between 0.84 and 2.84 MPa/mm for normal stress range of 0.25–
2 MPa, respectively. These normal reaction values have a good The difference of reaction modulus for both samples without
compatibility with experimental results presented by Mayer et al. and with gasket versus normal contact stress σ n is shown in Fig. 8.
[24]. The best fit trend lines for the relations of shear and normal As illustrated, the effect of gasket on the values of ks and kn
reaction coefficients with normal stress applied to the contact reduces with the increase in σ n . For instance, in the case of
surface of both samples with and without gasket were in the linear σ n ¼ 1 MPa, gasket presence causes in reduction of shear and
form as follows: normal reaction coefficients up to 6.32% and 0.88%, respectively
8 in comparison to without gasket condition.
>
< 1:4725σ n  0:1684 ðwithout gasket samplesÞ ð5Þ Comparison of the shear and normal reaction coefficients
R2 ¼ 0:9965
ks ¼ 1:4948σ  0:2504 2 ðwith gasket samplesÞ ð6Þ showed the maximum differences of 19.29% and 12.50% (for the
>
: n R ¼ 0:9953
normal stress of 0.25 MPa) and minimum differences of 0.36% and
ð5; 6Þ 0.23% (for the normal stress of 2 MPa), respectively. The presented
data in Fig. 8 proves that using gasket does not influence on shear
8
< 1:0638σ n þ 0:7307 ðwithout gasket samplesÞ ð7Þ and normal reaction coefficients, especially when the applied
R2 ¼ 0:9974
kn ¼ ðwith gasket samplesÞ ð8Þ normal stress is higher than 1.25 MPa.
: 1:1186σ n þ 0:6329 R2 ¼ 0:9957

ð7; 8Þ

where ks and kn are shear and normal reaction coefficients, 5. BCS model numerical implementation
respectively (in MPa/mm), σ n is normal stress applied to the
contact surface of the samples (MPa), and R2 is regression Purpose of this section is to evaluate the applicability of the
coefficient. obtained normal and shear reaction coefficients in numerical
The proposed equations can be efficiently used for contact modeling of Abuzar tunnel lining as an urban tunnel case study.
stiffness assessment which can be used in analytical modeling of The tunnel's geometry and geomechanical properties of the
the segmental liners. In this regard, the proposed reaction surrounding ground are shown in Fig. 1 and Table 1. For structural

Axial -70 Shear 15 Bending


9
Force Force Moment
10
6
-110 5
3
0
-150 -5 0

-10 -3
-190 -15 -6

Ie-min Ie -max Ie-min Ie-max Ie-min Ie-max

Horizontal 15.0
Vertical
Displacement Displacement 15.0
10.0 10.0
5.0 5.0
0.0 0.0
-5.0 -5.0
-10.0 -10.0
-15.0 -15.0

Ie- min Ie-max Ie - min Ie -max


Fig. 11. Axial and shear forces, bending moment, and vertical and horizontal displacements for minimum and maximum effective moments of inertia in EMI models.
A. Salemi et al. / International Journal of Rock Mechanics & Mining Sciences 77 (2015) 328–338 335

analyses using FEM, in this study, the concept of beam on elastic e. Eight models of beam and rotational spring (BRS) with the
foundation (BOEF) was used. flexural stiffness of 1–128 MN m/rad;
In this regard, gap elements were used with tangential and f. One model of beam contact springs (BCS) with shear and
radial stiffness based on Winkler's theory. Value of radial stiffness normal stiffness coefficients presented in this research.
was calculated using Eq. (9) while the tangential stiffness value
was assumed to be one-third of the radial one [30]:
Detailed configuration of the model geometry, loading condi-
K r ¼ Es =Rð1 þνs Þ ð9Þ tion, and the proposed contact model in this study is illustrated
in Fig. 9.
where Es and νs are elastic modulus and Poisson's ratio of the According to the model conditions specified in Table 1 and the
surrounding ground, respectively, and R is the tunnel radius. In existing 30 m overburden on the tunnel, the induced normal stress
total, 90 gap elements were considered to model the ground- on contact surface was obtained as 4.35 MPa by Eq. (4) and the
lining interaction. In this regard the radial and tangential stiffness shear and normal reaction coefficients were calculated as
coefficients of each gap element were calculated as 2.80 and ks ¼ 6:24 MPa=mm and kn ¼ 5:36 MPa=mm using Eqs. (5) and (7),
0.93 MN/m, respectively. respectively. To calculate the stiffness of shear and normal springs,
In this study, 22 analyses were carried out to compare the the reaction values were multiplied by the effective contact area of
situations of the existing models with the contact model presented segments (e.g. 1.2  0.15 m2 in this study). So, shear and normal
in this study; the list of the analyzed models is as follows: stiffness coefficients were obtained as ks ¼ 1123:76 MN=m and
kn ¼ 965:28 MN=m, respectively. In the proposed beam contact
a. One model of united lining (UL) without joints and with springs (BCS) model, contact points were modeled by linear
bending rigidity as EI; springs with the obtained stiffness coefficients. Internal forces
b. Nine models of reduced rigidity lining (RR) as ηEI using and displacements of structural members of the reduced rigidity
effective bending rigidity ratio (η) between 0.1 and 0.9; lining in the RR model varied with the ratio of effective bending
c. One model including hinged lining (HL); rigidity. For selecting the appropriate η value, nine states were
d. Two models with minimum and maximum effective moments modeled by nine ifferent η values ranging from 0.1 to 0.9. Results
of inertia (EMI) obtained from Eqs. (2) and (3); of these models are shown in Fig. 10.

Axial 20 Shear 15 Bending 4


Force 10
Force 10
Moment
2
0 5
-10 0 0
-20 -5
-2
-30 -10
-40 -15 -4

1 2 4 1 2 4 1 2 4
8 16 32 8 16 32 8 16 32
64 128 64 128 64 128

Horizontal 2.0 Vertical 10.0


Displacement Displacement
1.0 5.0

0.0 0.0

-1.0 -5.0

-2.0 -10.0

1 2 4 1 2 4
8 16 32 8 16 32
64 128 64 128

Fig. 12. Axial and shear forces, bending moment, and vertical and horizontal displacements with different values of kθ in BRS models.
336 A. Salemi et al. / International Journal of Rock Mechanics & Mining Sciences 77 (2015) 328–338

Results of the RR model illustrated in Fig. 10 show that the models with rotational stiffness of 1–128 MN m/rad were developed
values of internal forces and displacements increased in corre- using linear springs in joint locations, which are illustrated in Fig. 12.
spondence with η values. This increment was conservatively As shown in this figure in BRS models, variation of the
compared to that of other models. So, minimum value of effective rotational stiffness has not changed the axial force and vertical
bending rigidity ratio, i.e. η ¼ 0:1, was selected. displacement of the beam elements of the lining while the shear
For selecting the proper model from among the two EMI models force and bending moment have increased moderately with
with maximum effective moment of inertia (I e;max ¼ 0:11I) obtained increasing rotational stiffness coefficient. Consequently in corre-
from Eq. (2) and minimum effective moment of inertia spond to other developed models in technical literature, the
(I e;min ¼ 0:07I) calculated using Eq. (3), internal forces and displace- minimum and maximum values of rotational stiffness (i.e.
ments of the structural members of EMI model are given in Fig. 11. kθ ¼ 1 and 128 MN m=rad) were selected for comparison of the
Comparison of the results of two EMI models in Fig. 11 showed no results with other established models in the next section [21].
significant difference in terms of axial forces. Maximum axial force
of Ie-max model was about 1% more than that Ie-min model. Maximum
value of the shear force in the beam elements located in the first 5.1. Comparing results with the proposed model of the present study
and third quarters of lining was obtained using Ie-min, while in the
second and fourth quarters of lining; Ie-max would be used for Comparison of the results of UL, RR, HL, EMI, and BRS models
maximum results. Shear force of Ie-max model was higher than that with those of BCS model proves the close agreement of BCS and RR
of Ie-min model. Also, bending moment of Ie-max was greater than and HL (Fig. 13). Results of internal forces and displacements from
that of Ie-min and Ie-min model showed much more horizontal and these six models are shown in Tables 3 and 4, respectively. Table 3
vertical deformations than Ie-max model. Finally, in order to obtain contains maximum axial force and the corresponding shear force
conservative results as obtained in the united lining, the model with and bending moment and maximum bending moment and the
Ie-min was selected for comparison with the proposed model of corresponding axial and shear forces in the model. Table 4 con-
this study. tains displacements of roof, invert, and wall of the tunnel lining. It
Owing to extensive materials on rotational stiffness in the should be noted that, in these tables, different percentages of force
technical literature and in order to investigate the variation rate of and displacement values were calculated in proportion to the
internal forces and displacements of structural beam members, eight results of the proposed BCS model.

Axial 20 Shear 30 Bending


30
Force Force Moment
-30
15 15
-80
0 0
-130
-15 -15
-180

-230 -30 -30

UL RR UL RR UL RR
HL EMI HL EMI HL EMI
B-RS (Kθ=1) B-RS (Kθ=128) B-RS (Kθ=1) B-RS (Kθ=128) B-RS (Kθ=1) B-RS (Kθ=128)
B-CS B-CS B-CS

Horizontal 20.0 Vertical


20.0
Displacement Displacement
10.0 10.0

0.0 0.0

-10.0 -10.0

-20.0 -20.0

UL RRL UL RR
HL EMI HL EMI
B-RS (Kθ=1) B-RS (Kθ=128) B-RS (Kθ=1) B-RS (Kθ=128)
B-CS B-CS

Fig. 13. Axial and shear forces, bending moment, and vertical and horizontal displacements of aforementioned models of segmental tunnel lining.
A. Salemi et al. / International Journal of Rock Mechanics & Mining Sciences 77 (2015) 328–338 337

Table 3
Difference in the percentage of axial and shear forces and bending moment induced in structural members of 5 models with the proposed B-CS model.

Pmax (ton) M (ton m) V (ton) P (ton) Mmax (ton m) V (ton)

UL 166.47 25.71 1.63 166.47 25.71 1.63


 16.99%  382.36% 76.61%  111.10%  148.89% 27.56%
RR 147.70 7.36 1.63 147.70 7.36 1.63
 3.80%  38.09% 76.61%  87.29% 28.75% 27.56%
HL 154.64 3.86 7.89 90.52 7.84 2.20
 8.68% 27.58%  13.20%  14.79% 24.10% 2.22%
EMI 155.41 6.30 1.63 155.41 6.30 1.63
 9.22%  18.20% 76.61%  97.07% 39.01% 27.56%
B-RS (kθ ¼ 1 MN m=radÞ 30.80 1.07 2.97 12.84 3.13 7.47
78.35% 79.92% 57.39% 83.72% 69.70% 232.0 %
B-RS (kθ ¼ 128 MN m=radÞ 30.71 3.45 1.16 1.30 3.79 10.02
78.42% 35.27% 83.36% 98.35% 63.31% 345.3%
B-CS 142.29 5.33 6.97 78.86 10.33 2.25

Table 4
Difference in percentage of horizontal and vertical displacements of the structural members of 5 models with the proposed B-CS model.

Roof Wall Invert

H-Disp. (mm) V-Disp. (mm) H-Disp. (mm) V-Disp. (mm) H-Disp. (mm) V-Disp. (mm)

UL 1.97E  15  6.76E  03 6.77E 03 1.35E 05  3.68E  16 6.76E  03


814.03% 51.55% 36.70% 99.49% 109.13% 31.62%
RR 1.72E  16  1.29E  02 1.23E  02 1.20E  04  3.55E  17 1.29E  02
162.24% 7.21%  15.28% 95.46% 100.88%  30.96%
HL  5.08E  15  1.50E  02 1.42E  02 1.09E  03 1.28E  15 9.71E  03
 1742.51%  7.67%  32.65% 59.12% 68.25% 1.73%
EMI 4.27E  04  1.13E  02 1.49E  02 2.98E  05 4.27E  04 1.13E  02
– 19.00%  39.29% 98.88% –  14.32%
B-RS (kθ ¼ 1 MN m=radÞ 7.29E  17 4.04E  03 9.99E  04 6.57E  03 2.50E  16 5.15E  03
126.42% 128.97% 90.66%  147.31% 93.81% 47.88%
B-RS (kθ ¼ 128 MN m=radÞ 6.77E 16 4.02E  03 9.40E  04 6.52E  03 2.96E  16 5.09E  03
345.40% 128.82% 91.21%  145.43% 92.67% 48.49%
B-CS  2.76E  16  1.39E  02 1.07E 02 2.66E  03 4.04E  15 9.88E  03

Table 3 illustrates that BRS model was significantly different HL, and EMI models and that of BCS model belongs to RR model
from others. Maximum axial force of BRS model was 20% less than which is equal to 15.28% while the maximum difference is about
that of other models and was about 21% of BCS model. In addition, 39.29% that relates to EMI model. The minimum and maximum
shear force and bending moment of BRS model (kθ ¼ 1 MN m=rad) differences of the vertical displacement in the tunnel wall for RR,
were less than those of other models. Its maximum bending HL, and EMI models and those of BCS model are about 59.12% and
moment was also 50% less than others and about 30% of BCS 98.88% respect to HL and EMI models.
model. However, HL, RR, and EMI models had more correlation In addition, as shown in Fig. 13, axial and shear forces, bending
with the proposed BCS model, respectively. Maximum difference moment, and vertical and horizontal displacements of BCS model
of axial force among these three models and BCS model was 9.22% were in acceptable correlation with those of other models,
for EMI model and minimum difference was 3.80% for RR model. especially RR, HL, and EMI. In terms of segmental lining design,
Also, maximum difference of bending moment between these BCS model showed less axial force and bending moment than
three models and BCS model was 39% for EMI model and others, except in BRS. However, BCS demonstrated shear force
minimum difference was 24.1% for the HL model. values that were more similar to real cases, due to the existence of
As shown in Table 4, the best compatibility can be observed shear spring at contact points.
between displacement in RR, HL, and EMI models and also BCS
model. Comparing differences in horizontal and vertical roof
displacements of RR, HL, and EMI models with those calculated 6. Conclusions
for BCS model confirms that the minimum values belongs to RR
model. The minimum differences percentage between RR and BCS Stiffness of segmental lining at contact points is one of the
horizontal and vertical displacements at roof level are 169.24% and parameters that can affect the internal forces of lining. So, real
7.21%, respectively. The maximum difference of the vertical dis- behavior of contacts of segments must be considered in the
placement at roof level is 19%, which occurred for EMI model. analysis and design of segmental lining structure of tunnels. In
Among of RR, HL, and EMI models, the HL has the minimum and this research, some equations were presented for shear and
RR has the maximum differences of the horizontal and vertical normal reaction moduli at the contact points of segments with
displacements at tunnel invert level with those of BCS model. The grooved surfaces with or without gasket. Based on the results of
minimum differences percentages for BCS-HL models in horizontal experimental DST on the concrete samples of contact location, two
and vertical displacements are respectively 68.25% and 1.73% equations of shear reaction coefficient (ks ) and normal reaction
while the maximum differences percentage for BCS-RR models is coefficient (kn ) were derived in conjunction to normal stress for
30.96% in vertical displacement. The minimum difference percen- both samples with and without gasket. For validating the pro-
tage of the horizontal displacement of the tunnel wall between RR, posed contact model, 22 numerical models were developed for a
338 A. Salemi et al. / International Journal of Rock Mechanics & Mining Sciences 77 (2015) 328–338

water conveying tunnel in a case study framework. It was shown References


that in the range of 0.25–2 MPa for normal stress, results of DST on
the samples without gasket showed that shear stress threshold of [1] Ulusay R. SM for laboratory determination of direct shear strength. In: Hudson
sliding varied from 0.2 to 1.44 MPa and shear displacement JA, editor. The complete ISRM suggested methods for rock characterization.
Ankara: ISRM & ISRM Turkish National Group; 2007.
threshold of sliding was between 0.45 and 0.58 mm. In addition,
[2] Tang Y. The mechanism study of the staggering assembly of shield-driven
results of DST on the samples with gasket showed that shear stress tunnel. [M. Phil. thesis]. China: Department of Geotechnics, Tongji University;
threshold of sliding varied from 0.15 to 1.42 MPa and shear 1988.
displacement threshold of sliding was from 0.49 to 0.54 mm [3] Zhong X, Zhu W, Huang Z, Han Y. Effect of joint structure on joint stiffness for
shield tunnel lining. Tunnel Undergr Space Technol 2006;21:406–7.
according to the normal stress in the range of 0.25–2 MPa. [4] Morgan HD. A contribution to the analysis of stress in a circular tunnel.
In the case of the samples without gasket, values of the shear Géotechnique 1961;11(1):37–46.
reaction coefficient varied from 0.25 to 2.79 MPa/mm and normal [5] Peck RB, Hendron AJ, Mohraz B. State of the art of soft ground tunneling. In:
Proceedings of 1st rapid excavation and tunneling conference. Chicago: AIME;
reaction coefficients changed from 0.96 to 2.85 MPa/mm for the
1972. p. 259–85.
normal stress range of 0.25–2 MPa while in the samples with [6] JSCE. The design and construction of underground structures. Tokyo: Japan
gasket, shear reaction coefficient varied from 0.21 to 2.78 MPa/mm Society of Civil Engineers; 1977.
and normal reaction coefficient changed between 0.84 and [7] Ranken RE, Ghaboussi J, Hendron AJ. Analysis of ground-liner interaction for
tunnels. Report no. UMTA-IL-06-0043-78-3. Department of Civil Engineering,
2.84 MPa/mm for the normal stress range of 0.25–2 MPa.
University of Illinois at Urbana-Champaign; 1978.
The best fit trend for the relations of shear and normal reaction [8] Einstein HH, Schwartz CW. Simplified analysis for tunnel support. J Geotech
coefficients to normal stress applied to the contact surface in Eng Div ASCE 1979;105:499–518.
without gasket condition was linear regression equations by R2 of [9] Yuen CM. Rock–structure time interaction. [Ph.D. thesis]. London, Ontario:
Dept of Civil and Environmental Engineering, University of Western Ontario;
99% as ks ¼ 1:4725σ n  0:1684 and kn ¼ 1:0638σ n þ 0:7307, respec- 1979.
tively. In the mean while for contact surface with gasket the shear [10] Ogawa T. Elasto-plastic, thermo-mechanical and three-dimensional problems
and normal reaction coefficients were related to the normal stress in tunneling. London, Ontario: Department of Civil and Environmental
Engineering, University of Western Ontario; 1986.
induced with linear regression equations by R2 of 99% as
[11] Liu JH, Hou XY. Shield-driven tunnels. Beijing: China Railway Press; 1991. p.
ks ¼ 1:4948σ n  0:2504 and kn ¼ 1:1186σ n þ 0:6329, respectively. 152–303.
In overall, it was understood that using gasket caused reduction [12] Lee KM, Ge XW. The equivalence of a jointed shield-driven tunnel lining to a
in shear and normal reaction coefficients, which was decreased continuous ring structure. Can Geotech J 2001;38:461–83.
[13] Blom CBM. Design philosophy of concrete linings for tunnel in soft soils. [Ph.D.
with the increasing the normal stress. thesis]. Netherlands: Delft University of Technology; 2002.
Due to implementation of aforementioned equations for con- [14] El Naggar H, Hinchberger S, Lo KY. A closed-form solution for composite
tact points in numerical model of segmental lining of an urban tunnel linings in a homogenous infinite isotropic elastic medium. Can Geotech
J 2008;45:266–87.
tunnel and comparing the results with other existing contact
[15] Muir Wood AM. The circular tunnel in elastic ground. Geotechnique
models some interesting results were archived as follows: the 1975;25:115–27.
maximum axial force of BRS model was 20% of other models and [16] Hefny AM, Tan FC, Macalevery NF. Numerical study on the behavior of jointed
about 21% of the expected rate. In addition, shear force and tunnel lining. J Inst Civ Eng 2004;44:108–18.
[17] Hefny AM, Chua HC. An investigation into the behavior of jointed tunnel
bending moment of BRS model were less than those of other lining. Tunnel Undergr Space Technol 2006;21:428.
models. Maximum bending moment of BRS model was 50% of [18] Mashimo H, Ishimura T. Evaluation of the load on shield tunnel lining in
other models and about 30% of the expected value. However, HL, gravel. Tunnel Undergr Space Technol 2003;18:233–41.
[19] Lee KM, Hou XY, Ge XW, Tang Y. An analytical solution for a jointed shield-
RR, and EMI models had more correlation with the proposed BCS
driven tunnel lining. Int J Numer Anal Methods Geomech 2001;25:365–90.
model, respectively. Maximum difference of axial force between [20] Koyama Y. Present status and technology of shield tunnelling method in
these three models and BCS model was 9.22% for EMI model and Japan. Tunnel Undergr Space Technol 2003;18:145–59.
minimum difference was 3.80% for RR model. Also, maximum [21] Teachavorasinskun S, Chub-uppakarn T. Influence of segmental joints on
tunnel lining. Tunnel Undergr Space Technol 2010;25:490–4.
difference of bending moment among these three models and BCS [22] Arnau O, Molins C. Experimental and analytical study of the structural
model was 39% for EMI model and minimum difference was 24.1% response of segmental tunnel linings based on an in situ loading test—Part
for HL model. Minimum differences of the horizontal and vertical 2: Numerical simulation. Tunnel Undergr Space Technol 2011;26:778–88.
roof displacements of these three models and B-CS model were [23] Do NA, Dias D, Oreste PP, Djeran-Maigre I. 2D numerical investigation of
segmental tunnel lining behavior. Tunnel Undergr Space Technol
169.24% and 7.21%, respectively, which occurred in RR model. 2013;37:115–27.
Maximum difference of vertical roof displacement was 19%, which [24] Mayer PM, Libreros A, Hilber HM. Numerical modeling of cam-pocket coupling
happened for EMI model. Minimum differences of the horizontal system for concrete lining. Advances in construction material 2007, Part I.
Berlin: Springer; 2007. p. 57–65.
and vertical invert displacement of the three models and BCS
[25] Cavalaro SHP. Aspectos tecnológicos de túneles construidos con tuneladoray
model were 68.25% and 1.73%, respectively, related to HL model dovelas prefabricadas de hormigón. [Ph.D. thesis]. Universitat Politècnica de
and maximum difference of vertical invert displacement was Catalunya; 2009 [in Spanish].
30.96% in RR models. Minimum and maximum difference percen- [26] Molins C, Arnau O. Experimental and analytical study of the structural
response of segmental tunnel linings based on an in situ loading test–Part
tages of the horizontal displacement of wall between the three 1: Test configuration and execution. Tunnel Undergr Space Technol
models and BCS model were 15.28% and 39.29% for RR and EMI 2011;26:764–77.
models, respectively. Minimum and maximum differences of the [27] Cavalaro S, Aguado A. Packer behaviour under simple and coupled stresses.
vertical displacement of wall among the three models and BCS Tunnel Undergr Space Technol 2012;28:159–73.
[28] Do NA, Dias D, Oreste PP, Djeran-Maigre I. Three-dimensional numerical
model were 59.12% for HL and 98.88% for EMI models, respectively. simulation for mechanized tunnelling in soft ground: the influence of the joint
Altogether, results of the liner internal forces indicated that the pattern. Acta Geotech 2014;9:673–94.
proposed BCS model was more compatible with HL, RR, EMI, BRS, [29] Datwyler. Seals for tunnel construction—sealing profiles for modern tunneling
methods; 2014. 〈www.datwyler.com〉.
and UL models. Also, from the liner displacement point of view,
[30] Plizzari GA, Tiberti G. Steel fibers as reinforcement for precast tunnel
the proposed BCS model was more compatible with RR, HL, EMI, segments. In: Proceedings of the ITA-AITES 2006 world tunnel congress and
UL, and BRS models, respectively. the 32nd ITA General Assembly. Seoul; 2006. p. 438–9.

You might also like