Академический Документы
Профессиональный Документы
Культура Документы
Technical Note
a r t i c l e i n f o a b s t r a c t
Article history: The seismic analysis of underground structures requires a careful consideration of the important effect of
Received 19 February 2008 shear strains in the soil due to vertically propagating horizontal shear waves. These strains result in oval-
Received in revised form 13 October 2008 ing deformations of circular tunnels or racking deformations of rectangular tunnels. Closed-form solu-
Accepted 8 November 2008
tions as well as numerical analyses are used to characterize this soil-structure interaction problem.
Available online 25 December 2008
Many of these solutions assume full normal contact at the interface between the soil and tunnel lining.
This work describes a numerical finite element study of soil-circular tunnel lining interaction with con-
Keywords:
tact conditions that allow both limited slippage and separation to prevent development of potentially
Seismic analysis
Soil-tunnel interaction
unrealistic normal tensile and tangential forces at the interface. The analyses highlight the significant
Frictional contact limitations of widely used closed-form solutions in engineering practice. The finite element solutions
Underground structures demonstrate the need for realistic representation of the soil-tunnel interaction using numerical modeling
approaches.
Ó 2008 Elsevier Ltd. All rights reserved.
1. Introduction 1999 Izmit, Turkey earthquake, was the result of the intensity of
the seismic motions due to proximity of the seismic source to
While tunnels have generally performed better than above the tunnel.
ground structures during earthquakes, damage to some of these Three types of deformations characterize the response of under-
important lifeline structures during previous earthquake events ground structures to ground shaking: (1) axial compression and
highlights the need to account for seismic loading in the design extension; (2) longitudinal bending; and (3) ovaling/racking.
of underground structures. Ovaling/racking deformations have been the subject of many
Hashash et al. (2001) provide a review of damage to under- papers that describe closed-form solutions as well as numerical
ground structures and analytical approaches for seismic evalua- analysis approaches. For example, Kramer et al. (2007) describe de-
tions of these structures. In general, damage to tunnels is greatly tailed three-dimensional nonlinear ovaling analysis of a circular
reduced with increased overburden, and damage is greater in soils tunnel and Huo et al. (2006) developed closed-form solutions for
than in competent rock. Earthquake effects on underground struc- rectangular linings. By far the best known closed-form solutions
tures can be grouped into ground failure and ground shaking. to compute liner forces subjected to seismic racking were devel-
Ground failure constitutes conditions such as liquefaction, fault oped for circular tunnels by Wang (1993) and Penzien and Wu
displacement, and slope stability. (1998) and are widely used in engineering practice. Hashash
The collapse of the Daikai subway metro station in the 1995 et al. (2005) described the discrepancies between these two solu-
Hyogo-ken Nambu earthquake in Kobe, Japan, was the first such tions and used numerical analyses under the same assumptions
significant collapse of an urban underground structure due to to better understand the differences between the two the solutions
earthquake shaking, rather than ground instability. No seismic pro- and their causes. The closed-form solutions are limited to the fol-
visions were implemented in the 1962 design of this station as was lowing assumptions:
pointed out by Toshio et al. (1998) and the complete collapse of
this structure was due to loss of bearing capacity of the center 1. In the direction normal to the lining, soil and lining are fully
columns. Giannakou et al. (2005) suggests that the failure of connected.
Bolu tunnel in Turkey, which was under construction during 2. In the tangential direction, generally two cases are considered.
These two cases are best known as:
a. ‘‘full-slip” that assumes no tangential resistance transmitted
* Corresponding author. Tel.: +1 217 333 6986. from soil to the lining and
E-mail addresses: hassan@scsolutions.com (H. Sedarat), kozak@scsolutions.com
b. ‘‘no-slip” that assumes full connection between soil and the
(A. Kozak), hashash@illinois.edu (Y.M.A. Hashash), anoosh_shamsabadi@dot.ca.gov
(A. Shamsabadi), alex@scsolutions.com (A. Krimotat). lining.
0886-7798/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tust.2008.11.002
H. Sedarat et al. / Tunnelling and Underground Space Technology 24 (2009) 482–490 483
Fig. 1. Geometry and loading: (a) finite element model; (b) excavation load; (c) excavation force time function.
3. Tunnel is circular with uniform thickness and without any of tensile normal stresses at the soil-tunnel interface. A Coulomb
discontinuities. friction law is used to limit the tangential tractions. In this paper,
4. Soil and lining are massless and linear elastic. the terms ‘‘normal stresses” and ‘‘tangential tractions” are used
5. Plane strain conditions are postulated for soil and lining. to describe normal and tangential components of the soil-tunnel
6. The effect of construction sequence is not considered. interaction forces per unit area of the interface surface, respec-
tively. The term ‘‘total tractions” is also used as a reference to
These assumptions are often not representative of actual condi- vector summation of the normal and tangential components.
tions because The effect of frictional contact on the tunnel seismic response
is examined for a range of soil and lining stiffness and friction
1. Soil and lining are not integrally connected in the direction nor- coefficients. In all the cases, influence of gravity load and con-
mal to the lining. An exception would be when rock bolts or soil struction sequence on the state of stress of the tunnel is taken
reinforcement elements are used. into account. The results of the analyses show that the more real-
2. ‘‘Full-slip” or ‘‘no-slip” conditions are idealizations of real con- istic modeling procedure using frictional contact to represent the
tact of soil and lining. interface between the tunnel liner and surrounding soil results in
3. The shape of the lining is rarely circular or rectangular with uni- lining forces (due to seismically induced racking deformations)
form thickness. that are significantly different from those of the closed-form
4. Soil around the lining may yield. solutions. The study highlights the need for high fidelity tunnel
5. The construction sequence influences the state of stress at the and soil specific numerical modeling of seismic racking. It is
lining interface with the ground. not the intent of the study to develop alternative correlations.
Analysis of tunnels anchored to the surrounding soil with means
These deficiencies of the closed-form solutions are readily like rock bolts or soil reinforcement is outside the scope of this
overcome through the use of numerical analysis techniques pro- work.
vided that proper characterizations of the physical system are
implemented into the numerical models. Unfortunately, almost 2. Numerical modeling of tunnel ovaling with contact interface
all the numerical analyses presented in known literature incorpo-
rate the same assumptions as the closed-form solutions, and Fig. 1a shows the 2D finite element model used in this study
therefore, have the same limited applicability. In engineering and analyzed using the general purpose finite element program
practice there is sometime heavy reliance on closed form solu- ADINA (2005). The soil continuum and the tunnel are represented
tions (Wang, 1993; Penzien and Wu, 1998) without full recogni- with solid elements and beam elements of a unit width, respec-
tion of the limitations of these solutions. In this paper, a finite tively. The interface between soil and tunnel is modeled with fric-
element based approach with rigorous characterization of the tional contact. The coefficient of friction is varied to examine its
interface between soil and lining of circular tunnels through fric- effect on the tunnel response.
tional contact is proposed. Although only circular tunnel results It is important to have realistic initial stress conditions prior to
are reported in this paper, this does not result in any limitation application of racking deformations due to seismic loading when
of the proposed modeling approach and it is just a convenient the soil-tunnel interface is modeled with the frictional contact.
way to compare with existing closed-form solutions. The behav- The in situ state of stress is first established in the soil prior to
ior of the surrounding soil is assumed to be linear elastic, similar tunnel construction. The overburden pressure is applied at the
to the assumption in closed form solutions, in order to highlight top of the mesh as shown in Fig. 1a. The soil initial stress is spec-
the important role of modeling the soil-lining interface. In reality ified based on the assumed coefficient of lateral earth pressure K0.
soil behavior is highly nonlinear and should be represented in a In order to account for the soil stress relaxation and arching
numerical analysis. The contact condition prevents development during the tunnel excavation and construction, the widely used
484 H. Sedarat et al. / Tunnelling and Underground Space Technology 24 (2009) 482–490
Table 2
Seismic increment of lining sectional forces at soil shear strain of 0.5%; F = 143, t = 0.36m, K0 = 1.0.
Fig. 2. Lining total thrust at soil shear strain of 0.5%: (a) frictional contact (f = 1.0); (b) ‘‘no-slip” connection; displacement magnification factor = 20, F = 143, t = 0.36 m,
K0 = 1.0.
Fig. 3. Soil-tunnel interaction tractions at soil shear strain of 0.5%: (a) contact interface (f = 1.0); (b) ‘‘no-slip” connection; displacement magnification factor = 20, F = 143,
t = 0.36 m, K0 = 1.0.
Fig. 4. Soil maximum shear stress at soil shear strain of 0.5%: (a) contact interface (f = 1.0); (b) ‘‘no-slip” connection; F = 143, t = 0.36 m, K0 = 1.0.
486 H. Sedarat et al. / Tunnelling and Underground Space Technology 24 (2009) 482–490
Fig. 5. Contact tractions, lining total thrust and bending moment at soil shear strain of 0.5% under different friction coefficients: (a) f = 0 and (b) f = 1.0; F = 143, t = 0.36 m,
K0 = 1.0.
tangential tractions, the total tractions are much lower (Fig. 3a). In compressive thrust seismic increment versus soil shear strain is
the case of ‘‘no-slip” condition, total tractions at the soil-tunnel plotted for a set of friction coefficient values: 0, 0.5, 0.8, and 1.0.
interface are much larger with both compressive and tensile (push- These correspond to analyses with F = 143, t = 0.36m, k0 = 1.0. At
ing and pulling) normal components (Fig. 3b). Consequently, the early stages of ovaling, the soil-tunnel interface is locked in the
analysis with contact interface results in smaller regions in soil tangential direction which allows for a significant and rapid build
with high shear stresses compared with the ‘‘no-slip” condition up of thrust in the lining until the frictional limit resistance at
(Fig. 4). the interface is reached. This transition occurs at a higher value
of thrust for the higher value of the friction coefficient. Afterwards,
5. Effects of coefficient of friction, soil shear strain, and lining some local slippage between the lining and the soil develops. The
thickness on the tunnel response further increase in the thrust is due to the differential normal con-
tact stress. At this stage, the rate of increase in the thrust is the
Distributions of contact total tractions, lining total thrust and same for all friction coefficients (Fig. 6). In the course of ovaling
bending moment are compared for two values of friction coeffi- without friction (f = 0), thrust increases due to the normal contact
cient: 0 and 1.0 in Fig. 5a and b respectively. Other model param- pressure of the soil. In case of frictional contact (f > 0), the thrust
eters are F = 143, t = 0.36m, K0 = 1.0. At a soil shear strain of 0.5%, increment is developed due to the normal and tangential traction
the frictionless contact (f = 0) analysis results in contact stresses at the soil-tunnel interface until friction limit resistance is reached
that are normal to the tunnel (tangential component is zero) and and slippage occurs. From that moment on, further increase of the
relatively uniform: the deviation is between 28% and 14% from thrust is due to the same factor as in case of frictionless contact,
the mean value. Similarly the total thrust is relatively uniform namely, the normal contact pressure only.
along the tunnel circumference with a maximum deviation of Fig. 7 compares the seismic thrust increment from the numerical
±6%. The high friction coefficient (f = 1.0) analysis results in large analyses with the ‘‘full-slip” and ‘‘no-slip” closed-form solutions by
tangential traction with the maximum total traction 2.3 times lar-
ger than the maximum contact total traction for f = 0 analysis. De-
spite the fact that all normal contact components are compressive,
this contact total traction distribution results in tensile and com- f = 1.0
pressive thrust increments which lead to significantly non-uniform 2000
total thrust. The direction of the tangential tractions illustrates f = 0.8
Thrust (kN/m)
15975
7988
Thrust (kN/m)
2000
1000
0
Wang Full-Slip
Wang No-Slip
Wang No-Slip
Wang Full-Slip
Penzien No-Slip
Wang Full-Slip
Penzien No-Slip
Wang No-Slip
Penzien Full-Slip
Penzien No-Slip
Penzien Full-Slip
Penzien Full-Slip
Contact - f=1.0
Contact - f=1.0
Contact - f=1.0
Contact - f=0.0
Contact - f=0.0
Contact - f=0.5
Contact - f=0.8
Contact - f=0.5
Contact - f=0.8
Contact - f=0.0
Contact - f=0.5
Contact - f=0.8
Fig. 7. Seismic increment of lining thrust at different soil shear strains: 0.1%, 0.5%,
and 1.0%; F = 143, t = 0.36 m, K0 = 1.0.
Fig. 9. Seismic increment of lining thrust at soil shear strain of 0.5% under different
friction coefficients: 0, 0.5, 0.8, and 1.0; F = 3.23, t = 1.26 m, K0 = 1.0.
Table 3
Analyses’ matrix.
Case ID Lining radius, R (m) Lining thickness, t (m) Flexibility ratio, F Coefficient of friction, f Coefficient of lateral earth pressure, K0
2B3 01 5 0.36 14.3 0.0, 0.5, 0.8, 1.0 0.5, 1.0
2B3 5 0.36 143 0.0, 0.5, 0.8, 1.0 0.5, 1.0
2B3 10 5 0.36 1430 0.0, 0.5, 0.8, 1.0 0.5, 1.0
B3 01 5 1.26 0.32 0.0, 0.5, 0.8, 1.0 0.5, 1.0
B3 5 1.26 3.2 0.0, 0.5, 0.8, 1.0 0.5, 1.0
B3 10 5 1.26 32 0.0, 0.5, 0.8, 1.0 0.5, 1.0
488 H. Sedarat et al. / Tunnelling and Underground Space Technology 24 (2009) 482–490
Table 4
Maximum/minimum seismic increment of lining sectional forces at soil shear strain of 0.5% under different flexibility ratios: 14.3, 143, and 1430; t = 0.36m.
Flexibility ratio Force/moment* Closed-form solution Finite element solution with contact interface
Wang Penzien &Wu
Full-slip No-slip Full-slip No-slip K0 Coefficient of friction
0 0.5 0.8 1.0
14.3 max P 69 931 69 136 0.5 91 681 860 905
1.0 113 843 924 928
min P 69 931 69 136 0.5 72 628 819 858
1.0 80 813 913 922
V 69 136 0.5 130 125 119 112
1.0 122 120 108 107
M 345 345 345 341 0.5 299 276 270 266
1.0 301 263 264 264
143 max P 76 7988 76 152 0.5 217 811 1240 1550
1.0 231 1010 1545 1920
min P 76 7988 76 152 0.5 53 491 755 904
1.0 58 645 976 1160
V 76 152 0.5 132 131 132 132
1.0 130 131 132 132
M 380 380 380 379 0.5 332 324 314 309
1.0 334 320 308 303
1430 max P 77 36557 77 153 0.5 460 783 1024 1201
1.0 464 832 1104 1307
min P 77 36557 77 153 0.5 310 13 132 218
1.0 311 23 186 280
V 77 153 0.5 132 132 132 132
1.0 132 132 131 131
M 384 384 384 383 0.5 344 348 348 347
1.0 346 348 348 348
*
The lining forces and bending moment are designated as follows: P is the thrust increment (kN/m), V is shear force increment (kN/m), M is bending moment increment (kN-
m/m). Negative thrust increment indicates tension.
Table 5
Maximum/minimum seismic increment of lining sectional forces at soil shear strain of 0.5% under different flexibility ratios: 0.32, 3.2, and 32; t = 1.26 m.
Flexibility ratio Force/moment* Closed-form solution Finite element solution with contact interface
Wang Penzien &Wu
Full-slip No-slip Full-slip No-slip K0 Coefficient of friction
0 0.5 0.8 1.0
0.32 max P 571 1163 571 1034 0.5 589 1033 1160 1219
1.0 610 1169 1237 1243
min P 571 1163 571 1034 0.5 582 1150 1187 1225
1.0 594 1185 1241 1243
V 571 1034 0.5 1161 1110 1077 1060
1.0 1151 1068 1050 1049
M 2854 2854 2854 2585 0.5 2852 2694 2632 2611
1.0 2850 2631 2600 2598
3.2 max P 2271 9788 2271 4361 0.5 3416 3713 3958 4103
1.0 3155 3529 3805 3986
min P 2271 9788 2271 4361 0.5 202 2277 3437 4039
1.0 555 2858 4014 4637
V 2271 4361 0.5 3934 3926 3905 3907
1.0 3970 3963 3965 3948
M 11354 11354 11354 10903 0.5 9388 9206 9120 9086
1.0 9538 9319 9236 9205
32 max P 3234 66011 3234 6431 0.5 6593 6834 7066 7218
1.0 6406 6673 6909 7083
min P 3234 66011 3234 6431 0.5 1420 1748 3415 4200
1.0 1218 1985 3555 4455
V 3234 6431 0.5 5863 5950 5916 5840
1.0 5859 5934 5875 5879
M 16171 16171 16171 16077 0.5 13,267 13,023 12,926 12,882
1.0 13,271 13,012 12,920 12,872
*
The lining forces and bending moment are designated as follows: P is the thrust increment (kN/m), V is shear force increment (kN/m), M is bending moment increment (kN-
m/m). Negative thrust increment indicates tension.
thrust seismic increments for the tunnel lining thicknesses of increment increases with the flexibility ratio for a constant lining
0.36 m and 1.26 m, respectively, at a soil strain of 0.5% and stiffness. The thrust seismic increment increases as the thickness
K0 = 1.0. Except for the case of F = 1430, the lining thrust seismic of the lining increases for a constant soil stiffness. This can be
H. Sedarat et al. / Tunnelling and Underground Space Technology 24 (2009) 482–490 489
Table 6
Seismic increment of diagonal strain of the lining at soil shear strain of 0.5%.
Flexibility ratio Closed-form solution Finite element solution with frictional interface
Wang Penzien & Wu
Full-slip (%) No-slip (%) Full-slip (%) No-slip (%) K0 Coefficient of friction
0 (%) 0.5 (%) 0.8 (%) 1.0 (%)
14.3 0.63 0.63 0.63 0.62 0.5 0.59 0.55 0.53 0.53
1.0 0.58 0.53 0.52 0.52
143 0.69 0.69 0.69 0.69 0.5 0.64 0.63 0.63 0.63
1.0 0.66 0.63 0.63 0.62
1430 0.70 0.70 0.70 0.70 0.5 0.64 0.64 0.64 0.64
1.0 0.64 0.64 0.64 0.64
0.32 0.12 0.12 0.12 0.11 0.5 0.13 0.12 0.12 0.12
1.0 0.13 0.12 0.12 0.12
3.2 0.47 0.47 0.47 0.45 0.5 0.56 0.54 0.53 0.52
1.0 0.54 0.52 0.51 0.50
32 0.67 0.67 0.67 0.66 0.5 0.63 0.63 0.63 0.63
1.0 0.63 0.63 0.63 0.63
Flexibility Ratio The effect of coefficient of lateral earth pressure appears to be sig-
14.3 143 1430 nificant only for the 0.36 m thick lining and flexibility ratio equal to
143. The soil diametric strain values obtained from the finite ele-
36557
7988
2000
tions except for the case of the stiff lining (t = 1.26 m, F = 3.2).
The lining thrust seismic increments of the ‘‘full-slip” closed-
form solutions are comparable to those of the finite element
1000 model with the soil-tunnel contact interface only for the 1.26 m
thick lining and flexibility ratios equal to 0.32 and 3.2 (Fig. 11).
This is also true when Wang’s ‘‘no-slip” closed-form solution is
0 used for the soft soil (F = 14.3 in Fig. 10 and F = 0.32 in Fig. 11).
Wang Full-Slip
Penzien No-Slip
Wang Full-Slip
Penzien No-Slip
Penzien Full-Slip
Wang Full-Slip
Wang No-Slip
Penzien No-Slip
Wang No-Slip
Penzien Full-Slip
Wang No-Slip
Contact - f=1.0
Contact - f=1.0
Contact - f=1.0
Contact - f=0.0
Contact - f=0.0
Contact - f=0.5
Contact - f=0.8
Contact - f=0.5
Contact - f=0.8
Contact - f=0.0
Contact - f=0.5
Contact - f=0.8
cantly affect the level of the thrust in the tunnel lining, the tunnel
Thrust (kN/m)
Wang Full-Slip
Penzien No-Slip
Penzien No-Slip
Penzien Full-Slip
Wang Full-Slip
Penzien Full-Slip
Penzien Full-Slip
Wang No-Slip
Wang No-Slip
Penzien No-Slip
Wang No-Slip
Contact - f=1.0
Contact - f=1.0
Contact - f=1.0
Contact - f=0.0
Contact - f=0.0
Contact - f=0.5
Contact - f=0.8
Contact - f=0.5
Contact - f=0.8
Contact - f=0.0
Contact - f=0.5
Contact - f=0.8