Professional Documents
Culture Documents
ABSTRACT: Growing needs for modern transportation and utility networks have increased the demand for a
more extensive and elaborated use of underground space. The development of tunnelling projects under heavily
populated cities has been rapidly increased around the world during the last decades. Since tunnel construction
can significantly affect existing structures, construction methods have to provide maximum safety inside and
outside the tunnel. Conventional engineering models and codes for lining design simplify as e.g. by neglecting
longitudinal joints the interaction between rings through lateral joints. However,these days, in mechanized tun-
nelling we have to deal with a segmented lining system in contrast to a continuous tube (conventional approach)
that provide no information regarding lateral coupling and the longitudinal joints of the segmented lining sys-
tem. New design approaches for TBM tunnels, three-dimensional finite element analyses of a TBM tunnel,
where Ed. Zublin AG is involved as designer and contractor, are presented in this paper.
145
Table 1 details material parameters used for the FE tunnel lining is modelled with quadratic shape func-
simulations. tions elements (27 nodes). These higher-order ele-
ments allow for more accurate calculations than the
Soil Es c linear elements regarding momentum, normal and
MN kN kN shear forces. Details of numerical modelling of lining
m2
m3 m2
can be found in (Hilber & Raisch 1982).
Meletta,
highly 150 0.35 22 32.5 100 2.5
weathered
Meletta, 150 0.30 23 35 150 5
fresh Hinge element
Displacement (mm)
4.2 Hardening Soil Model
For the soil layers an elasto-plastic type of isotropic
hyperbolic material model was chosen, that was for-
mulated in the framework of hardening plasticity. An
important advantage of the Hardening Soil Model
compared to Mohr-Coulomb Model is the consider-
ation of the non-linear stress-strain relationship. The length of the beam
(m)
HSM considers the stress-dependency of stiffness
moduli and a hyperbolic stress-strain curve is used in- Fig. 1: Hinge element: numerical tests
stead of a bi-linear curve. Soil stiffness is described
more accurately by using three different input values:
triaxial loading stiffness, E50 , triaxial unloading re- 5.2 Jointed Rings
loading stiffness, Eur
ref , and the odometer loading stiff-
ness, Eoed . A detailed description of the model can be To study the bearing capacity of the system tunnel
found in the paper of (Schanz 1998). lining-soil, interface elements were used. The longi-
tudinal joints between single liner segments are con-
The material parameters used for modeling of the soil sidered as concrete hinges. Their behaviour was cali-
behavior with HSM are presented in Table 2. brated with test results from Leonhardt and Reimann
(Leonhardt & Reimann 1965). These interface hinge
ref ref ref
Soil E50 50 m Eur 50 ur elements allow non-linear behaviour. The rotation
MN kN kN resistance of the hinge is described by a retrac-
m2 - - m2 m2 - tion bending moment, which creates an eccentricity.
Meletta, The relation moment-rotation was empirically estab-
medium to lished from laboratory tests performed by Leonhardt /
completely 157 0.32 0.5 235 100 0.2 Reimann- see Heft 175 DAfStb. Numerical tests were
weathered conducted to study the behaviour of these hinge el-
Meletta, ements ( Figure 1). The non-linear behavior of the
150 0.30 23 35 150 5 hinge element is shown in Figure 1. For a value of
fresh
P = 10 kN, the beam responds within the elastic do-
Tab. 2: Soil properties adopted for HSM main. When increasing P to 56 KN, the the influence
of the hinge element becomes obvious and the hinge
element shows plastic behaviour.
5 NUMERICAL MODELLING For the numerical calculations the empirical formula-
5.1 Model Description tion was improved using the bilinear approximation
of Janssen (Janssen 1983). The dependence on rota-
The numerical solver used was three-dimensional fi- tion angle and related retraction bending moment m
nite element code of FEAT (Roddeman 1993). The can be determined and is illustrated in Figure 2.
146
0,4
0,35 Leonhardt/
Reimann
0,3 0.28
m=M/(N*b)
Fig. 2: m- dependency according to Leonhardt and Fig. 3: Simulation of longitudinal and circumferential
Janseen lining joints.
The equation for the elastic moment of the joint is: The circumferential joints used for Katzenberg tunnel
are modelled using thin continuum volume elements
M = c (1) (1cm thickness); their material behaviour follows the
Mohr -Coulomb criterion (Figure 4).
where M = moment, = rotation angle and c =
spring stiffness. The bending moment is determined
with the equation:
Z
M= n sdA = Km I0 (2)
147
35 250
As it was observed, the hard-pressed fibre plate
30
200
present a peak and a residual value of the coupling
force (Qres = 70%Qmax ). The kaubit coupling ele-
E-Modul [MPa]
25
20 150
148
The material parameters for the tunnel lining are: E = the tunnel are switched off (excavation) while the ele-
37000 MN/m2 , = 0.2 and = 25kN/m3 . The input ments of the tunnel lining are switched on (construc-
data for the coupling elements are: E = 27.7MN/m2 , tion).
= 0.2, = = 0 and c = 376 kN/m2 . An addi-
tional calculation for a smooth coupling system was t = 0,0 bis 1,0 t = 1,0 bis 2,0
performed, in order to check the plausibility of the re-
sults. For this purpose the cohesion of the coupling el-
ement was considered zero. The normal bedding was G
verified with a quasi- 2D -FE model that focussed on
the surrounding soil / rock (Figure 7). To determine
the subgrade modulus a continuum ring with a length
of 2 m was simulated, where the ring stiffness was
reduced acc. to an approach by Muir-Wood (Muir- t = 2,0 bis 3,0 t = 3,0 bis 4,0
Wood 1975).
10 m
7,7 m
7 RESULTS
10,8 m
The model mesh extends horizontally ca. 50 m from Figure 9 illustrates the difference between both mod-
the tunnel axes to the edge, which amounts to four els. Unloading, in this case resulting from the exca-
times tunnel diameter; vertically the model extends vation of the tunnel, is associated with a soil stiffness
25.5 m below the tunnel base. increase, which is simulated by both materials laws.
However, the stiffness increase produced by MCM is
To ensure that the boundaries do not significantly af- lower than by HSM, which led to unrealistic deforma-
fect the results, the sedimentary cover was limited to tions within the soil and the lining annulus. The cal-
44.8 m. For soil layers exceeding that realm, a sur- culations show that HSM reproduces the stress-strain
face load of 790.7 kN/m2 was applied. The numerical relationship more realistically.
modelling of tunnel construction required four cal-
culation steps (see Figure 8). The unloading process The HSM calculations were validated with simula-
of the rock mass is obtained by reducing the forces tions of triaxial tests where the secant stiffness of a
around the excavation by 50%, which amounts to a soil subjected to similar load conditions as around the
volume loss of 0.48%.During step one, the layers are tunnel was calculated. The stiffness corresponding to
activated and soil/rock properties are allocated. Dis- a 50 % maximum shear strength (565 MN/m2 ) dif-
placements obtained during this step are disregarded. fers slightly from the stiffness produced after the first
During step 2, the forces in the excavation plane are stages of tunnel construction Es1 = 430 MN/m2 (see
reduced and during step 3 the soil elements inside Fig. 10).
149
-2500 -2500
Ring 5 Ring 6
-2000 -2000
sigyy [kN/m^2]
sigyy [kN/m^2]
-1500 -1500 w [mm] w [mm]
-1000 -1000
10.50
10.50
11.82
11.82
13.57
13.57
17.70
15.59
9.07
15.58
9.07
17.71
12.70
-500
11.26
-500
16.53
9.98
11.26
16.52
9.98
15.41
14.11
12.70
15.40
14.11
0 0
0 -0.001 -0.002 -0.003 -0.004 -0.005 0 -0.001 -0.002 -0.003 -0.004 -0.005
eptyy [-] eptyy [-] Fig. 12: Displacements of two coupled adjacent rings
Fig. 9: Stress-strain diagram for soil behavior mod-
elled with MCM ( left) and HSM (right) calculation M N Q u
-2500 -2000
Model kN/m2 kN/m kN/m mm
qr = 2158 kN/m continuum
-2000 -1500 Muir-Wood 523.1 1845 279.7 14.1
(one ring)
1 [kN/m]
1 [kN/m]
-1500 -1000
E S2 = 689 MN/m FEM
-1000
0,5 qr = 1079 kN/m
coupled rings 553.1 1828 331 9.68
-500
E 50 = 565 MN/m
E S1 = 430 MN/m
jointed
-500
0 -0.01 -0.02 -0.03 -0.04 -0.05
0
0 -0.001 -0.002 -0.003 -0.004 -0.005
one ring 467.2 1950 246.9 17.47
1 [-] 1 [-]
FEM
562.9 1921 315.9 10.33
Fig. 10: Comparison of results from laboratory tests coupled rings
with FE calculations using HSM Tab. 3: Calculations results for phase fourth
7.2 Coupled Rings
Results from calculations of tunnel lining without continuum and coupled rings for the Katzenberg Tun-
joints and coupled rings with joints are compared. nel lining. Note, that that the bending moment for the
Figure 11 shows the deformed shape of a meso-scale jointed coupled rings is largest (562,9 kNm/m). The
model without joints owing to swelling pressure . reason for this difference is the introduction of a cou-
pling system within the longitudinal joint.
8 SUMMARY
The research presented is summarized as follows:
150
Janssen, P. (1983). Tragverhalten von Tunnelausbauten mit
Gelenktuebbings. Bericht Nr. 83-41 aus dem Institut
fuer Statik der technischen Universitaet Braunschweig.
Leonhardt, F. & Reimann, H. (1965). Betongelenke, Ver-
suchsbericht, Vorschlaege zur Bemessung und kon-
struktiven Ausbildung. Dafstb. Verlag Ernst und Sohn,
Berlin.
Muir-Wood, A. (1975). The circular tunnel in elastic
ground. In Geotechnique 25, pp. 115127.
Roddeman, D. (1993). Tochnog professional, version 2.0 -
user manual. Print by FEAT.
Schanz, T. (1998). Zur Modellierung des mechanischen
Verhaltens von Reibungsmaterialien. Ph. D. thesis, Mit-
teilungshefte des Instituts fuer Geotechnik, Heft 45.
STUVA (1996). Eignungspruefungen 4. Roehre Elbtunnel,
Versuche zur Verdrehsteifigkeit. Koeln.
151