Académique Documents
Professionnel Documents
Culture Documents
091]
The paper illustrates a three-dimensional finite-element analysis aimed at back-predicting the response
of a multi-storey reinforced concrete building underpassed by a metro tunnel. The study refers to the
case of the Milan metro line 5, recently built in coarse-grained materials using an earth pressure
balance machine, for which settlement measurements along ground and building sections were
available. The soil behaviour is modelled using an advanced constitutive model that, when combined
with a proper simulation of the excavation process, proves to realistically reproduce the subsidence
profiles recorded in free-field conditions. The building is found to modify the deformative pattern at
the ground surface in relation to its stiffness and weight, reducing the differential settlements as
compared to those calculated under free-field conditions. Results of the numerical simulation carried
out, including the model of the building schematised in detail, are found to be in good agreement
with the monitoring data. They thus indirectly confirm the satisfactory performance of the adopted
numerical approach, which takes into account a unique model of the soil, the tunnel and the building
that is, the key ingredients of this interaction problem. Further analyses are also carried out
modelling the building, adopting different levels of detail. The results highlight that, for the case
under study, the simplified approach based on the equivalent plate schematisation is inadequate to
capture the real displacement field. The overall behaviour of the system proves to be mainly
influenced by the buried portion of the building, including its foundation elements, which plays an
essential role in the interaction mechanism.
KEYWORDS: finite-element modelling; monitoring; settlement; soil/structure interaction; tunnels
INTRODUCTION
The construction of underground tunnels in urban areas
often requires excavation works to be carried out in close
proximity to residential buildings, cultural heritage monuments and underground services. The ability to predict the
tunnelling-induced settlements and the associated impact on
pre-existing structures represents a key aspect to estimate
potential damages and to design protective measures, when
needed (e.g. Mair, 2008; Amorosi et al., 2012; Puzrin et al.,
2012; Rampello et al., 2012).
Soil deformation and structural response are often assumed to be decoupled, so that the building damage is
typically predicted based on free-field settlement profiles
(Peck, 1969; Burland & Wroth, 1974; Burland et al., 1977;
OReilly & New, 1982; Boscardin & Cording, 1989; Burland, 1995). However, such a simplified approach disregards
the influence of the structure stiffness (Potts & Addenbrooke, 1997; Franzius et al., 2006) and the role of its
weight (Franzius et al., 2004), often leading to rather
conservative solutions in terms of estimated differential
settlements and, consequently, of induced damage intensity.
In the last few years two-dimensional (2D) and threedimensional (3D) numerical approaches have been developed
to overcome such limitations. 2D numerical studies were
proposed by Liu et al. (2000) with reference to surface
masonry structures, focusing on the effect of facade weight,
stiffness and position with respect to the tunnel axis. Alongside the same topic, Amorosi et al. (2014) back-analysed the
23
24
97 m
Lotto Station
57 m
50 m
GS_1
GS_2
43 m
GS_3
GS_4
50 m
46 m
GS_5
GS_6
Scale: 50 m
Left tunnel
Right tunnel
6m
25
Tun
n
7
el m 0 m
ode
lled
por
tion
Portello Station
Fig. 1. Milan metro-line 5: plan view of the segment of the route between Lotto and Portello stations
137 m
193 m
248 m
180 m
197 m
D 669 m
GS_6
Ground benchmarks
Building targets
Scale: 10 m
Left tunnel
Right tunnel
L2
L1
L3
L4
L5
T1
M
building
R1
R2
R3
T2
R4
R5
B 12 m
T3
15 m
70
Tun
n
el m
L 30 m
ode
lled
por
tion
(a)
25
(b)
Fig. 4. (a) General view of main left side facade and (b) detail of the garage zone on the right longitudinal side of the building
26
13 m
13 m
13 m
II
III
13 m
13 m
IV
31 m
19 m
VI
30 m
300 m
VII
42 m
VIII
52 m
64 m
Structure above
ground level
19 m
31 m
51 m
50 m
Garage below
ground level
g. l. m a. s. l.
GS_1
GS_2
0
R
R
SG
G, S(L)
CD_2
z: m
5
S, G(L)
S, G(L)
10
10
SG
15
S, G
L(S)
w.t._2013
w.t._2007
20
15
Tunnel axis
Tunnel invert
25
G(S)
Tunnel crown
20
Horizontal scale: 10 m
L(S)
24
G, S(L)
30
Fig. 6. Soil conditions along the examined portion of the route, depth of tunnel and position of ground monitoring sections
Computed:
Gravelly sand
27
Ground monitoring
Transversal settlement profiles due to the excavation of
the first tunnel are shown in Fig. 8. They refer to fully
developed settlements achieved when the tunnel face was at
a sufficient distance from the monitoring sections.
The maximum settlement induced by the excavation never
exceeds 7 mm. Measurements are sufficiently well fitted by a
Gaussian distribution curve (Peck, 1969) for K values in the
range 0 .40 .45, with the exception of the left-most points of
each section, whose values are under-predicted by the empirical relation (Fig. 8), probably due to the influence of nearby
surface structures located along the examined route (see Fig.
1). The above interpolation allowed back-evaluation of the
corresponding volume loss, VL (%), which varies from 0 .3%
to 0 .38%, with an average value equal to 0 .33%, indicating a
well-performing EPB excavation, and consistent with similar
observations reported in the literature (Leblais & Bochon,
1991; Ata, 1996; Mair, 1996; Mair & Taylor, 1997).
The values of maximum settlement, K parameter and
volume loss for all the considered sections are summarised
in Table 1. The table also reports the different depths of the
tunnel axis, z0, at each location and the recording date of
the analysed settlement measurements.
The evolution of settlement above the tunnel centre-line is
presented in Fig. 9 as a function of the face distance for the
two sections, GS_3 and GS_6, for which measurements of
vertical displacements close to the tunnel face (i.e. 1 m)
were available. The tunnel face settlement, Sv,f, at these
locations is equal to about 1 mm, indicating a satisfactory
face support during the excavation process.
Measurements were interpreted at each location by the
cumulative Gaussian probability curve (Attewell & Woodman, 1982) in order to define the longitudinal settlement
trough, assuming the volume loss and K values reported in
Table 1 and considering the longitudinal inflection point, iy,
equal to the transversal one, ix. As shown in Fig. 9, face
settlements are best fitted by the translated Gaussian cumulative curve (Mair & Taylor, 1997), obtained by equating the
ratio Sv,f /Sv,max to the measured one. This translated profile,
however, is not able to capture the further evolution of
settlements, predicting the achievement of steady-state conditions well before what is observed in situ.
Sandy silt
10
Depth, z: m
S_1
15
20
L
25
S_2
30
200
400
600
Small-strain shear modulus, G0: MPa
800
Structural monitoring
Vertical displacements of the building were gathered during
tunnelling by monitoring targets located along the base of its
three sides, identified by capital letters L, R and T and a
sequential number (see Fig. 3). The target relative distance and
their distances from the tunnel axis are listed in Tables 2 and 3.
As expected, the vertical displacements mainly increase
with the excavation advancement, as shown in Figs 10(a)
10(c), where the evolution at each monitoring point is
displayed at different dates. Final measurements, gathered
when the distance of the right tunnel face from section
GS_6 was about 50 m, that is about 8D, range from 4 .7 mm
to 6 .5 mm along the longitudinal left facade, from 3 .5 mm
to 6 .6 mm along the longitudinal right facade and from
4 .6 mm to 5 .7 mm along the transversal side.
These plots make it possible to highlight the effect of the
distance from the tunnel axis on settlement: the closer the
tunnel axis (i.e. points L1, L2, L3 along the longitudinal left
side, R4, R5 along the longitudinal right side and T2, T3
along the transversal one), the higher the settlement.
The progressive settlement response of the building
evolves during tunnelling, this being particularly evident
along its longitudinal sides: the structure, in fact, is characterised by a hogging-type mode of deformation when the
tunnel-boring machine (TBM) face is located in correspon-
28
Settlement, Sv: mm
Settlement, Sv: mm
8
40
20
0
20
Distance from tunnel axis, x: m
(a)
20
0
20
Distance from tunnel axis, x: m
(b)
40
20
0
20
Distance from tunnel axis, x: m
(d)
40
20
0
20
Distance from tunnel axis, x: m
(f)
40
Settlement, Sv: mm
Settlement, Sv: mm
8
40
40
8
40
20
0
20
Distance from tunnel axis, x: m
(c)
8
40
40
Settlement, Sv: mm
Settlement, Sv: mm
8
40
20
0
20
Distance from tunnel axis, x: m
(e)
40
8
40
Measurements:
Transversal troughs:
GS_1
GS_3
GS_5
Empirical
GS_2
GS_4
GS_6
Computed
Fig. 8. Transversal settlement troughs: measurements and best-fitting Gaussian curves (the computed profile for
ground section GS_6 is also shown)
Monitoring
sections
NUMERICAL MODEL
Geometry and finite-element discretisation
Different numerical models were set up to simulate the
tunnel excavation under free-field conditions and in the
presence of the building, the latter being modelled with
different levels of detail. In the interaction analyses, for the
sake of simplicity, the presence of the nearby structures was
neglected.
The mesh employed in the present study is shown in Fig.
11(a): it represents a soil volume 68 m wide, 30 m high and
100 m long. These dimensions were selected to minimise the
influence of the boundary conditions on the computed
results. The numerical model adopted for the free-field
analysis is constituted by 84 125 nodes; this number increases to 120 952 in the interaction analysis with the
complete structural model. Nodes at the bottom of the mesh
GS_1
GS_2
GS_3
GS_4
GS_5
GS_6
Sv,max:
mm
5 .6
5 .2
6 .0
5 .5
6 .5
5 .5
VL: %
0 .45
0 .42
0 .45
0 .42
0 .40
0 .45
0 .38
0 .30
0 .37
0 .30
0 .31
0 .34
z0: m
Date
20 .0 16 December 2012
19 .0 18 December 2012
18 .0 20 December 2012
17 .0
8 January 2013
16 .0 10 January 2013
15 .0 15 January 2013
dence with the middle of the building (measurements recorded on 11 January 2013), while the deformative pattern
evolves in a sagging-type mode after the tunnel passage (i.e.
from 12 January 2013 on).
In general, no evidence of damage was detected on this
Settlement, Sv: mm
Monitoring points
40
20
0
20
40
60
80
Distance from tunnel face, y: m
(a)
100
L1
L2
L3
L4
L5
R1
R2
R3
R4
R5
T1
T2
T3
8
60
120
Settlement, Sv: mm
29
2
4
6
8
60
40
20
0
20
40
60
80
Distance from tunnel face, y: m
(b)
Measurements:
100
120
Longitudinal troughs:
Original
GS_3
Translated
GS_6
Computed
Relative distance: m
5 .54
6 .80
7 .09
8 .44
7 .18
7 .32
6 .97
7 .33
4 .84
6 .38
L1L2
L2L3
L3L4
L4L5
R1R2
R2R3
R3R4
R4R5
T1T2
T2T3
L2
L3
L4
L5
R1
Settlement, Sv: mm
Settlement, Sv: mm
2
4
6
8
R2
R3
R4
R5
T1
(a)
2
4
6
8
T2
T3
Settlement, Sv: mm
L1
(b)
2
4
6
8
(c)
Recording date:
7 January 2013
10 January 2013
13 January 2013
8 January 2013
11 January 2013
14 January 2013
9 January 2013
12 January 2013
15 January 2013
Fig. 10. Structural vertical displacements recorded in correspondence with the monitoring targets: (a) L1L5; (b) R1R5;
(c) T1T3. Settlements recorded at targets T1T3 before 12 January 2013 are equal to zero
30
y
x
10
0m
68 m
(a)
y
z
(b)
Fig. 11. (a) Sketch of the mesh used in the interaction analysis.
(b) A detail of the longitudinal section on the finite-element model
is also shown
Name
Values
Gravelly sand soil S_1 Sandy silt soil L Gravelly sand soil S_2
: kN/m3
20
17 .5
20
0
33
0
5
26
0
0
33
0
0 .4
48 000
48 000
144 000
0 .2
250 000
0 .0001
0 .85
54 250
54 250
162 750
0 .25
155 000
0 .0002
0 .4
58 944
58 944
176 832
0 .2
307 000
0 .0001
100
0 .455
0 .9
0
0
100
0 .562
0 .9
0
0
100
0 .455
0 .9
0
0
Failure parameters:
c9: kPa
9: degrees
: degrees
Effective cohesion
Effective friction angle
Dilatancy angle
Stiffness parameters:
m
E9ref
50 : kPa
E9ref
oed : kPa
E9ref
ur : kPa
ur
ref
G0 : kPa
0 .7
Other parameters:
pref: kPa
K nc
0
Rf
tension
cincrement: kPa/m
31
Shield
Face pressure
40 kPa
VI
14 m slice
14 m 14 m
80 kPa
98 m
VII
120 kPa
VII
160 kPa
Shield
Lining
0 .03
75
0 .25
210
0 .3
25
0 .15
35
IV
200 kPa
III
240 kPa
II
I
32
8
I
12
30
20
II VII III
IV
10
0
10
Distance from tunnel axis, x: m
(a)
20
30
II
Settlement, Sv: mm
Settlement, Sv: mm
FF_VL 032%
STR_VL 036%
12
30
20
III
VI
10
0
10
20
30
Distance from tunnel axis, y: m
(b)
40
50
Settlement, Sv: mm
Settlement, Sv: mm
12
33
10
15
20
Longitudinal left side, l: m
(a)
25
12
30
10
15
20
Longitudinal right side, l: m
(b)
25
30
Settlement, Sv: mm
Measurements_middle
Measurements_end
FF (point M)
STR (point M)
FF
STR
12
6
9
Transversal right side, l: m
(c)
12
15
Fig. 15. Comparison of monitored and computed settlements of FF and STR analyses on the longitudinal (a) left and
(b) right sides and (c) on the transversal right side of the building
STR analysis
STR analysis
1400
1400
800
600
Normal force, N: kN
Settlement, Sv: mm
1000
800
600
10
15
20
Building left side, l: m
(a)
25
0
30
200
1400
STRwcb analysis
1400
10
15
20
Building right side, l: m
(b)
25
0
30
STRwcb analysis
1200
1000
800
600
400
Normal force, N: kN
1200
Settlement, Sv: mm
Normal force, N: kN
400
400
200
1000
800
600
Settlement, Sv: mm
Normal force, N: kN
1000
Settlement, Sv: mm
1200
1200
200
400
0
10
15
20
Building left side, l: m
(c)
Column position
0
30
25
Sv
200
N before tunnelling
10
15
20
Building right side, l: m
(d)
25
0
30
N after tunnelling
Fig. 16. STR and STRwcb analyses: normal compression force and settlement values at the base of the columns
on the left (a, c) and right (b, d) longitudinal sides of the building
34
CONCLUSIONS
This paper presents the results of a coupled geotechnical
and structural numerical study aimed at investigating the
response of a multi-storey building affected by tunnellinginduced settlements. This topic is currently relevant as the
ever-increasing demand for urban space leads to underground developments, which represent a possible cause of
structural damage for surface structures. The study, conducted using a three-dimensional finite-element code, refers
to the recent construction of the metro-line 5 in Milan
(Italy).
The soil behaviour is described by a non-linear elastoplastic constitutive model (termed hardening soil with
small-strain stiffness) calibrated with reference to in-situ
tests. The main aspects of the excavation process are reproduced in the 3D numerical simulation of the EPB tunnelling.
0
Settlement, Sv: mm
Settlement, Sv: mm
12
10
15
20
Longitudinal left side, l: m
(a)
25
10
15
20
Longitudinal right side, l: m
(b)
25
30
Settlement, Sv: mm
Settlement, Sv: mm
12
30
12
6
9
Transversal left side, l: m
(c)
Measurements_end
12
15
12
6
9
Transversal right side, l: m
(d)
STR
STRwcb
STRW
STR*
12
15
Fig. 17. Comparison of monitored and computed final settlements on the longitudinal (a) left and (b) right sides, and
(d) on the transversal right side of the building. (c) The computed settlement profiles on the transversal left side are
also compared
35
G0 1 aj=0.7 j
(3)
Eur
2(1 ur )
(4)
ACKNOWLEDGEMENTS
Financial support provided by Astaldi S.p.A. in the person
of Eng. Enrico Campa is gratefully acknowledged.
Special thanks go to Eng. Giuseppe Colombo of Milano
Serravalle Milano Tangenziali S.p.A. (formerly Astaldi
S.p.A.) and to Eng. Davide Fraccaroli and Eng. Alessandro
Caffaro of Astaldi S.p.A. for providing the monitoring data
and for the technical support during the site activity.
bw 0.175(h hw )0 4 d w
(5)
36
n
X
(Ec A)slab
(8)
(Ec J )building
n
X
(Ec J )slab Ec
n
X
(J slab Aslab H 2m )
(9)
where n is the reference level of the building; Aslab and Jslab are the
cross-sectional area and the second moment of area of the slab at
each level, respectively; and Hm is the vertical distance between the
slabs and the structures neutral axes (the latter assumed to be
located in correspondence with the structures centroid).
The building foundation system was neglected in this simplified
approach (Franzius et al., 2006). The computed axial and bending
stiffness for each slab are reported in Table 6, together with the
thickness and Hm values.
The input parameters of the plate element used in the finiteelement analysis were then evaluated as
s
12(Ec J )building
tfe
(10)
(Ec A)building
Efe
(Ec A)building
tfe
(11)
with tfe and Efe being the equivalent thickness and the Youngs
modulus, respectively.
The unit volume weight of the plate element, which is equal to
2 .92 kN/m3, was calculated as the ratio between the total building
weight (excluding the weight of the retaining walls modelled in
STR analysis, equal to about 2500 kN) and the plate volume
(B 3 L 3 tfe).
Table 6. Stiffness properties of the slabs at each level
Level, n
Basement floor
Ground floor
First floor
Second floor
Third floor
Fourth floor
Fifth floor
Sixth floor
Seventh floor
Eighth floor
Ninth floor
Slab
thickness: m
Hm: m
EcAslab: kN
EcJslab: kNm2
0 .26
0 .26
0 .22
0 .22
0 .22
0 .22
0 .22
0 .22
0 .22
0 .22
0 .22
15 .70
13 .20
9 .00
5 .80
2 .60
0 .60
3 .80
7 .00
10 .20
13 .40
16 .60
1 .95 3 108
1 .95 3 108
1 .65 3 108
1 .65 3 108
1 .65 3 108
1 .65 3 108
1 .65 3 108
1 .65 3 108
1 .65 3 108
1 .65 3 108
1 .65 3 108
1 .10 3 106
1 .10 3 106
6 .66 3 105
6 .66 3 105
6 .66 3 105
6 .66 3 105
6 .66 3 105
6 .66 3 105
6 .66 3 105
6 .66 3 105
6 .66 3 105
NOTATION
Aslab
Aw
B
bw
c9
37