Technical Paper On Design
Technical Paper On Design
Technical Paper On Design
ABSTRACT: The paper addresses precast tunnel segments for the new Line 9 of Barcelona Metro. After
the experimental determination of the material properties, non-linear analyses allowed to study the structural
behavior of the tunnel segments under TBM thrust jack forces during construction process. The paper focuses
its attention to the advantages coming from the use of Steel Fiber Reinforced Concrete (SFRC) in terms of crack
control and bearing capacity. Finally, Ultra High Performance Concrete (UHPC) has been investigated to study
if this material is suitable for the application in tunnel segments.
1461
Mechanics (NLFM; Hillerborg et al., 1976) with a
smeared crack approach. The numerical model has
been validated by using the experimental results
of splitting tests performed on SFRC specimens
(Schntgen and Erdem, 2003). During the validation,
different concrete crack models have been tested and
compared.
2 MATERIALS
Fibers were used with the volume fractions (Vf ) equal 4,0
Gf
w w
3,0
Experimental
Fracture properties of FRC were determined by 2,0
FEA DIANA Smeared
using eight notched beams (150 150 600 mm) 1,0
0,0
tested under 4-point bending according to the Italian 0,0 0,1 0,2 0,3
Standard (UNI, 2003; Fig. 2). In addition, 4 beams CTODm [mm]
of plain concrete were used as reference specimens.
The slump of the fresh concrete was always greater Figure 3. Experimental and numerical results obtained from
than 150 mm. Specimens were stored in a fog room SFRC beams and the bilinear law adopted.
(R.H. > 95%;T = 20 2 C) until 24 hours before test-
ing. The mechanical properties of concrete (average bridge action of concrete between microcracks while
values), as determined after about 60 days of cur- the second branch represents the residual stress due to
ing are the following: tensile strength (determined fiber bridging (Fig. 3). The numerical analyses were
from cylinders having = 80 mm and L = 240 mm) performed by assuming both a discrete crack and a
fct = 4.10 MPa, compressive strength (from cubes with smeared crack approach with Diana ver. 9.1 (2005);
a side of 150 mm) fc,cube = 64.1 MPa andYoungs mod- the latter was also used for the numerical simulations
ulus (from compression test on cylinders = 80 mm, of the tunnel segments.
L = 240 mm), Ec = 37000 MPa. The material parameters identified from the bend-
Fracture tests were carried out with a closed-loop ing tests are the following (see Fig. 3): fct = 4.10 MPa,
hydraulic testing machine by using the Crack Mouth 1 = 2.134 MPa, w1 = 0.023 mm, wcr = 5.35 mm. The
Opening Displacement (CMOD) as control parameter, best-fitting numerical curves obtained with Diana are
which was measured by means of a clip gauge posi- compared with the experimental ones in Figure 3.
tioned astride a notch at midspan, having a depth of Firstly, numerical analyses of tunnel segments were
45 mm. Additional Linear Variable Differential Trans- carried out with different amounts of steel fibers;
formers (LVDTs) were used to measure the Crack secondly, the response with Ultra High Performance
Tip Opening Displacement (CTOD) and the vertical Concrete (UHPC) has been investigated. The post-
displacement at midspan and under the load points cracking properties of concrete with different fibers
(Fig. 2). contents were assumed starting from the bi-linear law
Inverse analyses of the bending tests, based on experimentally determined from concrete C50/60
NLFM (Hillerborg et al., 1976), allowed to deter- with 45 kg/m3 of steel fibers.
mine the best-fitting post-cracking law (-w) for The stressstrain relations and the values of UHPC
the SFRC adopted in the present research (Roelfstra were obtained according to den Hollander (2006). The
and Wittmann 1986). The Youngs modulus (Ec ) was stress-strain diagram for UHPC is given in Figure 4.
the one experimentally measured from the cylinders UHPC was made with two types of fibers: micro
while the Poisson ratio () was assumed equal to 0.2. and macro (l = 50 mm) fibers; the latter govern the
The softening law was approximated as bilinear where ultimate tensile strain. The characteristic values of the
the first steeper branch can be associated with the stressstrain diagram are given in Table 1.
1462
Figure 4. Stressstrain diagram for UHPC with strain
hardening (den Hollander, 2006).
Table 1. Material properties for the adopted UHPC mixture. Figure 5. Spalling cracks observed during construction of
Line 9 of the Barcelona Metro.
Material
property Symbol Value The research work presented herein focuses on the
last topic. In order to verify the influence of steel
Compressive strength fck 200 MPa fibers on the crack development in segmental linings
Design comp. strength fc1 120 MPa during construction, several non-linear analyses were
Design tensile strength t1 fct1 8 MPa
carried out.
Design tensile strength t2 fct2 9 MPa
Design tensile strength t3 fct3 5 MPa Based on the material properties described in 2,
Yield compressive strain c1 1.75 the total strain rotating crack model provided by
Ultimate compressive strain cu 3.00 DIANA was adopted for FE simulations of the SFRC
Ultimate tensile strain ctu 10.00 tunnel segments. The numerical analyses were car-
Specific weight c 28 kN/m3 ried out by adopting a 3D solid model with 5184 first
Youngs modulus Ec 55000 Mpa order hexahedral elements (eight nodes brick elements
HX24L). The constitutive law for concrete under com-
pression was assumed according to Thorenfeldt (1987)
that is a suitable model for a concrete grade C50/60.
3 DESIGN ASPECTS Moreover, the increase of compressive strength due to
lateral confinement has been implemented according
An open question for the construction companies and to Selby and Vecchio (1993).
the designers concerns the reinforcement for these pre- During excavation process, 30 TBM actuators act
cast elements. Generally, reinforcement is designed on the Barcelona Metro lining. The average service
according to the design actions on the tunnel segments, load applied by a single actuator is 3 MN.The load con-
resulting from segment transportation, placing process figuration used in Barcelona is a combination of the
and soil pressure in the final state. French and Japanese jack configuration. In practice,
In particular, during construction, the thrust forces two pairs of thrust jacks act on each tunnel segments
and the grout pressure are the most critical factors. whereas four bearing pads are positioned in the rear
Cracks often appears in the tunnel lining in the phase face of the tunnel segments (ring joints).
in which these forces occur. Some examples of cracks One single segment with his two pairs of actuators
that have been observed in Line 9 of the Barcelona was considered and its boundary conditions (i.e. pres-
Metro are shown in Figure 5. Possible causes of these ence of adjacent segments) were simulated by elastic
cracks could be: eccentricity, inclination of the thrust springs whose stiffness was previously calibrated.
jacks. Also a number of phenomena due to the trum- Noticing that also ring joints have been simulated by
pet shape as the torsional deformation or an uneven using no-tension elastic springs positioned on the
support of the ring joint cause these cracks. bearing pad surfaces (Fig. 6).
It is desirable to mitigate or reduce these cracks Figure 6 shows the loading scheme when thrust
as much as possible since they determine a loose of jacks are exactly placed in position and the support
quality and leakage and high repair costs. Cracking of the ring joint is uniform.
phenomena can be limited by introducing a change Figure 7 shows the numerical results in terms of load
in tunnel design by using, for example, an oppor- (from the actuators) versus the average longitudinal
tune configurations of the thrust jacks and supports. displacement under the loading surfaces. The segment
Alternatively, they can be reduced by using FRC. collapses at a total load of 30.7 MN; since the service
1463
Figure 6. Scheme of a segment with the load from the two
pairs of jacks, four rear bearing pad supports and spring
representing the stiffness of the neighbor segments.
35
30
Figure 8. Scheme of the measurements of the splitting
25 cracks.
Total load [mN]
20
15 35
Service Load
10 30
Normal loading configuration
25
Total load [mN]
0 20
0 0,5 1 1,5 2 2,5 3 3,5
15
Average displacement under load surfaces [mm]
Normal loading situation, 20 cm
10 Normal loading situation, 30 cm
Figure 7. Load displacement curve with the nominal load Normal loading situation, 40 cm
5
scheme.
0
0 0,2 0,4 0,6 0,8 1,0
Displacement [mm]
load is 12 MN, the safety factor is 2.58. The compres-
sive strength of concrete is almost reached in longi- Figure 9. Development of splitting crack width under
tudinal stresses under loading surfaces (c = 55 MPa, the load.
fc = 58 MPa). As shown in Figure 7, the first cracks
appear between the loading surfaces, due to spalling, However, a number of irregularities can occur in
at approximately 6.0 MN (0.5 times service load). The practice. As a consequence, the ring joint may not be
spalling cracks develop with an increase of load and plane and the thrust jacks may be not exactly on place.
penetrate more deeply in the segment. For this reason, a possible eccentric placement of the
Splitting cracks appear under the loading surfaces thrust jack, its orientation and an uneven support in the
at 17.5 MN, due to splitting stresses in radial direc- ring joint have been investigated. In particular, numer-
tion (r ). In the classical elastic solution, the depth of ical results of and eccentric load are presented herein
the disturbed zone is approximately the width of the in comparison with normal load conditions.
structure that means, in this case, the thickness of the Eccentric placement of thrust jack in radial direc-
lining. tion has been considered because it will increase the
After cracking, the splitting stresses decrease in the spalling stresses and can damage the segment. The
cracked zone and are redistributed over a larger area eccentric placement has been modeled in FE analy-
with respect to the classical elastic solution. Because sis as a triangular pressure distribution on the loading
of the concrete toughness provided by fibers, cracks surface. This provokes an eccentricity of 37.3 mm
develop in a stable way. (Fig. 10).
Figure 9 shows the development of the splitting When the above mentioned eccentricity is present
crack width under the load. The graph represents the outward (Fig. 10b) the ultimate load decreases sig-
relative displacements of two nodes: one located inside nificantly. In fact, the numerical analyses show a
of segment and the other on the outside. Three pairs maximum load of 23.6 MN, that corresponds to a
of nodes are taken on 20, 30 an 40 cm along the safety factor of 1.97 (Fig. 11). Spalling cracks ini-
longitudinal direction from the TBM side of the seg- tially appear between the thrust jacks; they start with a
ment (Fig. 8). The graph evidences that there exists a load of 5.0 MN and significantly influence the segment
redistribution of stresses after cracking. collapse.
1464
35
Inside tunne l Inside tunne l
30
25
35
Figure 13. Development of splitting stresses under the load.
30
25
Total load [mN]
20
15
Service Load
10
Eccentricity outside
5
Normal loading configuration
0
0 0,5 1 1,5 2 2,5 3 3,5
Average displacement under load surfaces [mm]
Figure 14. Cracks between the jack loads observed in the
Line 9 of the Barcelona Metro and FE crack patterns.
Figure 11. Loaddisplacement curve with and without
eccentricity.
35
30
25
Total load [mN]
Because of the eccentricity, the segment tilts out- Figure 15. Load-displacement curves for a segment with
different amounts of steel fibers.
ward providing a non-smooth support. Consequently,
a bending moment occurs, as shown by the 3D model
in Figure 12. numerical simulations. In the Barcelona Metro tun-
Splitting stresses initiate at an earlier load level nel these cracks appeared when 9.3 MN were applied
(11.4 MN instead of 17.5 MN without eccentricity); by the jacks (corresponding to 0.78 times the target
this is probably due to higher local stresses in the load) whereas the numerical crack patterns refers to a
longitudinal direction. The development of these split- load of 11.4 MN. Looking at the similarities between
ting cracks under the load is schematically shown in the numerical and the actual cracks in the tunnel,
Figure 13. eccentricity could explain the presence of these cracks.
The cracks between the loads develop over the total The normal load situation has been simulated by
depth of the segment due to the increased bending considering also SFRC with a different amount of steel
moment for the outward tilting. It can be noticed that fibers. The adopted fiber contents are 15, 30, 45, 60
similar deeply penetrating cracks between the thrust and 75 kg/m3 . The numerically obtained ultimate load
jacks have been observed in the Line 9 of the Barcelona ranges from 24.3 MN (for 15 kg/m3 ) to 33.5 MN (for
Metro. Figure 14 exhibits the location of the cracks of 75 kg/m3 ; Fig. 15). The constitutive post-cracking laws
the segment as well as the crack patterns obtained by for the different fiber contents are shown in Figure 15.
1465
5 40 4 CONCLUDING REMARKS
Crack width at 24.3 mN [mm]
35
4
The present paper shows results of a numerical study of
25
precast tunnel segments for the Line 9 of the Barcelona
3
20
Metro. The numerical model shows reasonable results
2
in terms of bearing capacity and crack development in
15
a segment during TBM operations.
Crack width 10
1
Ultimate load
Eccentric placement of the thrust jack is one of the
5
possible explanations of longitudinal cracks that are
0 0 often present the in tunnel linings, including the one
0 15 30 45 60 75 90
of the Barcelona Metro. In fact, the presence of eccen-
Amount of steel fibres [kg/m3]
tricities is one of the possible unfavorable conditions
Figure 16. Influence of the amount of steel fibers on the
that causes higher stresses in the lining. In this load-
ultimate load and on the maximum crack width. ing situation, it has been demonstrated that the ulti-
mate load reduces significantly and larger cracks are
90 present.
80 The bearing capacity of the segment under jack load
70 increases with the fiber content up to a maximum value
Total load [mN]
1466
In BEFIB 2004, Proceedings of the 6th RILEM Sympo- Schntgen B. 2003. Design of Precast Steel Fibre Rein-
sium on FRC, Varenna (Italy), September 2022, RILEM forced Tunnel Elements, in Test and Design Methods for
PRO 39, pp. 41156. Steel Fibre Reinforced Concrete Background and Expe-
Kooiman, A.G. 1996. Steel fibre reinforced concrete in riences, Proc. of the RILEM TC 162-TDF Workshop,
shield tunnel linings, a state of art. Stevin report 25.5- Bochum (Germany), March 2021, 145152.
96-10. Delft University of Technology. Selby, R. G., & Vecchio, F. J. 1993. Three-dimensional Con-
Plizzari G.A. & Cominoli L. 2005. Numerical Simulations stitutive Relations for Reinforced Concrete. Tech. Rep.
of SFRC Precast Tunnel Segments. In Proceedings of 9302, Univ. Toronto, Civil Eng., Toronto, Canada.
ITA-AITES 2005, Underground space use: Analysis of Thorenfeldt, E., Tomaszewicz, A. & Jensen, J.J. 1987.
the past and lessons for the future, Erdem Y. and Solak T. Mechanical properties of high-strength concrete and
Eds., Istanbul (Turkey), May 712, pp. 11051111. applications in design. In Proc. Symp. Utilization of
Plizzari G.A. & Tiberti G. 2006. Steel fibers as a reinforce- High-Strength Concrete (Stavanger, Norway).
ment for precast tunnel segments, in the World Tunnel UNI 11039. 2003. Steel fibre reinforced concrete Part I:
Congress ITA AITES 2006, Seoul Corea, April 2227. Definitions, classification specification and conformity
Roelfstra, P.E. & Wittmann, F. H. 1986. Numerical method Part II: Test method for measuring first crack strength and
to link strain softening with failure of concrete. In ductility indexes. Italian Board for Standardization (UNI).
Fracture Toughness and Fracture Energy of Concrete, Waal R.G.A. de. 1999. Steel fibre reinforced tunnel seg-
Wittmann F.H. Ed., Elsevier, Amsterdam, 163175. ments, Delft, ISBN 90-407-1965-9, Delft.
1467