You are on page 1of 6

Numerical Methods in Geotechnical Engineering – Schweiger (ed.

)
© 2006 Taylor & Francis Group, London, ISBN 0-415-40822-9

Morphological effects on settlements induced by shallow tunnelling

D. Segato & G. Scarpelli


Technical University of the Marche Region, Ancona, Italy

ABSTRACT: Stress and strain fields near the ground surface are strongly influenced by ground morphological
conditions, so that sometimes, in the case of shallow tunnel design, numerical predictions of the induced subsidence
are required to account for an initial stress which may not be simply geostatic. In this paper, this problem is studied
with reference to a real tunnel excavated in a stiff clay deposit under the most intensively urbanized area of the
city of Ancona (Italy). The results of the monitoring show that, displacements at the ground surface caused by
tunnel excavation can be only explained if soil morphology is taken into account. In this particular problem, the
morphological effect arise as a consequence of the sloping that characterize the ground surface at the beginning
of the tunnel. A numerical, finite element model of the tunnel problem is presented to clarify such a behavior
through parametric analysis.

1 INTRODUCTION study). The result is that, while when dealing with


support design the 2D simplification may be conser-
The correct formulation of boundary conditions, is vative, in settlement curve prediction the plane strain
a key parameter in analyzing any engineering prob- hypothesis may lead to an unsafe evaluation.
lem. Numerical analysis of geotechnical problems is Bearing in mind the points mentioned above, in the
generally worked out using the plane strain condition following, a simplified (with reference to geometries,
hypothesis, since software supporting three dimen- materials and boundary conditions) FEM model, is
sional models are still expensive in time and sometimes presented in order to enhance the comprehension of
difficult to interpret. Dealing with tunnel analysis, the monitoring data from S. Martino Tunnel. Furthermore,
plane strain hypothesis is generally acceptable, in the model is used for a sensitivity analysis with refer-
the cross plane, only far from the advancing front where ence to the problem of the morphological effect.
the strain field, due to the presence of the soil core,
becomes three dimensional.
The current method of analysis is to “force” the 2D
2 ANALYSIS OF CASE RECORD
numerical solution to take into account the presence
of the front face using the convergence-confinement
2.1 Description of San Martino tunnel
method introduced by Panet & Guenot (1982). At the
excavation boundary a pressure reproducing the “at The San Martino tunnel has been driven through
rest” condition is applied, which is then reduced a stiff clay deposit beneath one of the main hills
through the deconfinement coefficient. For shallow where the Ancona old city has expanded. It connect
tunnels excavated in urban areas, care must be taken the port area (Vanvitelli building) with the city center
in defining boundary conditions because the plane (S. Martino square). Figure 1 gives a schematic layout
strain hypothesis may be inappropriate even far from of the tunnel which is characterized by an about 100 m2
the face: due to low covers and short lengths, the stress excavation section and a 536 m length. The liner
state is highly influenced by topographic effects. The consists of a primary steel rib and jet grout lining at
presence of the slope standing above the portal allows the advancing face and a secondary reinforced con-
a three dimensional state in the first part of the tunnel crete internal layer cast in place at about a 2 diameter
before the excavation takes place which will result in distance.
a preliminary strain field different from that expected The tunnel ends were already present to a depth of
in a 2D cross section analysis. 50 m and as such they served as bomb shelter during
The more the soil element is close to the slope, the the Second War years. The former excavation section
more this effect becomes important (resulting in a set- in these zones was about 85 m2 and was supported by
tlement up to three times greater in the present case a brick lining.

299
20,0

Vertical settlement: mm
25,0

30,0

35,0
CL section
sectionFG
FG
40,0 CL section
sectionHG
HG
CL section
sectionIG
IG
section 1Dfrom
section 1D fromCL
CL
45,0
0 50 100 150 200 250
Distance from the edge of the slope (longitudinal): m

Figure 1. Tunnel layout. Figure 2. Longitudinal settlement profiles recorded along


tunnel center line (CL) and one diameter aside.

The “Vanvitelli” side of the tunnel (left on fig. 1),


has been studied in the present paper. It must be 30 IG IG
HGHG
FG FG
observed however that, except for the slope facing the
sea which dipped 45° upward from the portal, the tunnel 20
underlay, at an average height of 50 m, an intensely Vertical settlements: mm 10
urbanized area.
Before the excavation started, a surface monitoring 0

has been set and some holes along the tunnel axis were -10
instrumented with both magnetic extensometers and
-20
slope indicators for vertical and horizontal displace-
ments respectively. Piezometers were also installed -30 Vs = 1.9%
FG progr 0+225
for pore pressure measurements. In the following some -40
Vs = 3.0%
HG progr 0+175 Vs = 2.9%
result of topographic monitoring will be examined so
IG progr 0+120
as to identify the influence of morphological effect on -50
-80 -40 0 40
the results.
Distance from tunnel axis (orthogonal): m

Figure 3. Measured and theoretical settlement troughs taken


2.2 Results of field monitoring parallel to the slope edge, at different cross sections.
As the excavation progressed, surface monitoring
showed the development of a zone along the tunnel
center line were high vertical settlement were observed 300 10,00
compared to adjacent sections. Such a behavior was not Pore pressures at CL array FG:
tunnel depth 5,00
vertical settlements: mm
275
in agreement with internal convergence measurements, Settlement at CL array FG 0,00
250
Pressure: KPa

which in turn did not show unexpected volume loss, and -5,00
neither was observed any local change in mechanical 225
-10,00
properties of the mass. 200
-15,00
Figure 2 provides a plot of longitudinal settlements 175
-20,00
recorded after the excavation completion. Transverse 150 -25,00
settlement profiles taken parallel to the top slope edge
125 -30,00
and illustrated in figure 3 show a typical Gaussian 06/10/03 03/02/04 02/06/04 30/09/04 28/01/05 28/05/05 25/09/05
trend; the theoretical settlement pattern based on the Date
methods proposed by Peck (1969) and Mair et al.
(1993) and matching each profile with the proper Figure 4. Pore pressure and measured vertical settlements
cover, indicate a subsidence volume increasing while at FG center line (CL) on FG.
approaching the slope. Plotting pore pressure and
vertical settlement vs tunnel advancement allows an
appreciation of how, the subsidence response is that
of a drained continuum (fig. 4). the tunnel lining was water tight, for the whole obser-
This can be explained as the consequence of the vation period the pressure didn’t come back to the
macro structural features of the clayey deposits crossed former value as the hydrographic basin was of small
by the tunnel. It can be also noted that, even though extension.

300
Table 1. Summary of general soil properties.

Soil property Stiff clay

Unit weight (kN/m3) 20


OCR (at gravity loading stage) 3
K0 1.18
G0 (MPa) 150
E⬘ (MPa) 360 Figure 5. Mean stress contours and principal stress orien-
E⬘50 ref (MPa) 40 tation along the tunnel axis.
E⬘oed ref (MPa) 30
c⬘ (kPa) 80
␾⬘ (deg) 26 4 DETAILS OF THE ANALYSIS
⌿ (deg) 5
4.1 Preliminary morphological considerations
Since the free surface constituted by the slope in the
3 SOIL PROPERTIES FOR ANCONA first part of the tunnel, is a boundary condition for the
STIFF CLAY principal stress flow, the shallow layer stress pattern
will arrange parallel to this plane.
From the geomechanics point of view, in the considered Owing to this fact, horizontal stresses in the driving
mass portion, the two main soil strata can be identified direction taken at different distances from the slope
from the ground surface: top edge will be variable and generally different from
that expected in K0 conditions (which are instead
– S: clayey sandy silt about 8 m thick;
reached far from the slope). It must be remembered
– C: stiff overconsolidated clay deposit, typically
that in this case, the final configuration of the principal
found in the area dating back to the Pliocene period.
stress field will be also dependent on the mass stress
Due to the aim of the present study, it appeared to be history. This will affect obviously the amount the K0
useful to adopt models which have to be simple, but value in the transverse direction, and will induce a
still are able to give a correct representation of the rotation of principal stress thus modifying vertical
phenomena which are present in the soil mass. To and horizontal (in the driving direction) stresses in the
reach this goal, the whole mass has been assumed as mass (fig. 5). When in the considered case the plane
having the same mechanical properties of the C layer. strain excavation at different distances from the slope
For sake of simplicity, in the FE analysis, soil behavior edge is analyzed, the variability of the out of plane
has been modeled through the use of the so called stress (and strain) field can not be taken into account
Hard Soil model, an elastic/plastic relationship with leading to an identical stress state and thus same trans-
deviatoric isotropic hardening which takes account for verse settlement trough.
plastic deformations (both volumetric and distortional)
and is implemented in the FE codes PLAXIS 7.2 and
3D Tunnel (Vermeer & Brinkgreve, 1998). In the HS 4.2 Plane strain conditions
model, the stress strain relationship is hyperbolic and In order to match the 2D settlement trough, the FG
the adopted tangent stiffness E⬘ (which controls the array data were taken as reference and a simple FE
material stiffness at each strain level) has been related problem model has been created which is showed in
to the shear stiffness at small strains G0; this value has figure 6.
been chosen, on the base of a wide bibliography avail- The tunnel excavation is simulated in a 2D green-
able on such materials (Scarpelli, 1995), equal to field FE analysis. The parameters utilized for the HS
150 MPa. From E⬘, the reference stiffness values to be model are reported in table 1, and formerly discussed.
entered in the model are defined through FE simulation A circular excavation section has been considered with
of triaxial CID tests and the comparison with tests a 10 m diameter (D), an H vs. D ratio of 5 have been
results on real soil samples; it is pointed out that, being considered as cover height. Since the pore pressure
interested in small strain behavior, special care has drop recorded during monitoring can be associated
been used in reproducing the initial part of q-␧ curves. to a generalized phreatic lowering, at the tunnel axis,
Typically, stiff clays in the Ancona geological con- (thus not affecting the shape of the settlement profile)
text have medium/high OC rates in the first 20 ÷ 25 m, for sake of simplicity the presence of the water was
decreasing with depth; in the following analysis, to neglected.
represent the behavior of the soil element which is Construction sequence was modeled using the
involved in subsidence phenomena, an OCR ⫽ 3 at convergence-confinement method. During the first
the gravity loading stage has been taken for the whole phase of the analysis, a ␭ value equal to 0.5 has been
mass which corresponds a 1.18 value for K0 OC (tab. 1). used and in the following phase a circular lining

301
10D

D 10D

H Figure 8. FE mesh for 3D analysis of San Martino tunnel


excavation.

Distance from the edge of the slope (longitudinal): z/H


0 0.4 0.8 1.2 1.6 2 2.4 2.8 3.2 3.6 4
0,00
Figure 6. Plane strain model for FE analysis.
-0,01

Vertical settlements: m
0,00
-0,02
-5,00
Vertical settlements: mm

-0,03
-10,00

-15,00 -0,04

-20,00 -0,05

measured at FG array -25,00


-0,06
2D computed -30,00
Figure 9. Evolution of the longitudinal settlement profile
-35,00 with tunnel progress for greenfield conditions with H ⫽ 50 m
-100 -80 -60 -40 -20 0 20 40
and K0 ⫽ 1.18.
Distance from the axis: m

Figure 7. Comparison between vertical settlements meas- improve significantly the computed settlement pro-
ured at array FG and computed in plain strain conditions. file has not yet been defined.

4.3 Three dimensional model


modeled by an elastic beam has been activated bringing
the ␭ value up to a value of 1.0. Being the analysis in The 3D FE model shown in figure 8 has been devel-
drained conditions, the volume (per running meter) of oped using PLAXIS 3D Tunnel code by extruding the
the surface settlement trough equal to that measured former 2D model and excavating the tunnel in the z
in the real case was taken as matching criteria. direction. The 45° slope has been created after the
A volume loss higher than that expected was calcu- gravity load phase by removing soil elements. In the
lated, which may be partially due to the inability of the following, coordinates will be referred to the origin of
hyperbolic relationship to reproduce the plastic soft- z axis set below the slope top edge and normalized to
ening of the stiff clay experienced by soil close to the the height of the hill (50 m). The analysis has been car-
tunnel profile and partially to the reduced excavation ried on in drained conditions as for the previous case.
area used in the model. As the analysis purpose is to When introducing the third dimension in the prob-
study the strain field far from the excavation area and lem, the convergence-confinement method in plane
being the mass mainly in pre failure conditions except strain conditions has to be substituted by the step by
a small area around the excavation, such an approxi- step modeling sequence (Katzenbach & Breth, 1981).
mation was considered acceptable. The Calculated and Since the excavation sequence is not volume loss
measured settlement trough are illustrated in figure 7. controlled, an excavation length of about 20 m (2D)
The inaccuracy of the matching between measured with the contemporaneous activation of the lining has
and computed settlement trough when the initial stress been chosen, which gives in plane strain conditions,
profile is characterized by high values of the coefficient the same radial convergence as in the 2D case. As the
of lateral earth pressure at rest K0, has been discussed tunnel heading is advancing, the longitudinal settlement
by several authors (Addembroke et al., 1997; Gunn, profile on the top of the slope develops as shown in
1993; Simpson et al., 1996; Lee & NG, 1996; Franzius, figure 9. Different curves are given for every advance-
Potts & Burland, 2005) but a feature which is able to ment step.

302
0 0,5 1 1,5 2 2,5 3 3,5 4 0 0,5 1 1,5 2 2,5 3 3,5 4
1 -0,5

Normalized settlements: Sv/Sv2D


Normalized settlements: Sv/Sv2D

-1
1,5

-1,5
2
45° slope -2 45° slope K0 = 1.18
2,5 35° slope
45° slope K0 = 0.873
27° slope -2,5
45° slope K0 = 0.562 (NC)
3 45° slope K0 = 1.18 friction ang 35°
-3

3,5 -3,5
Distance from the edge of the slope (longitudinal): z/H Distance from the edge of the slope (longitudinal): z/H

Figure 10. Longitudinal normalized settlement profiles for Figure 11. Longitudinal normalized settlement profiles for
different slope inclination. different initial stress states.

The figure demonstrates that for the reference case, Results of this set of analysis are shown in figure 11.
in addition to the cumulative error curve shape across Again the extension of the influence area is still the
the tunnel face (which moves forward as the heading same and inside it high K0 values emphasize the slope
advances), a permanent influence zone 2H wide close effect being the NC case unaffected. The case with
to the slope edge is produced where settlements are higher friction angle resulted in less influence in terms
amplified. In the 3D model this influence results in of maximum settlement ratio because of the smaller
an amplification of the gaussian shape settlement extension of the plastic zone around the excavation.
curve approaching the slope top edge. It must be noted When an additional surface is introduced, the stress
that similar behavior may be obtained if a volume loss rotation reduces again the influence of the main slope.
is applied to the excavation section in a single step;
this implies that the observed surface settlement trough
may be attributed to the lack of confinement in the 4.5 Synthesis of the results
slope influence area. The former results show an invariability in the exten-
sion of the influence zone regardless the initial stress
4.4 Parametric analysis
state and the inclination of the slope.
A series of parametric analyses to investigate the If only the excess settlements with respect to the
influence of various parameters was carried out. The plane strain conditions is considered, it is possible to
first one illustrates the influence of slope steepness. plot the normalized profile over the typical settlement
Three cases are shown in figure 10 where for an ini- profiles for spandrel type excavations for cantiliver
tial K0 ⫽ 1.18, 45°, 35° and 27° sloping hill sides are walls. In figure 12, the computed excess profile for
considered and vertical settlement are normalized to the care which have resulted sensitive to the influence
the value recorded in plane strain conditions. zone are overlapped to the curves proposed by Hsieh &
The figure highlights two effects: in this stress state Ou (1998). It is interesting to note that the extension
conditions the effect of the slope on the settlement of the primary influence zone is the same as that pre-
through is negligible for slopes under 27°. Furthermore, dicted by the tunnel FEM analysis.
the extension of the influence area doesn’t change being
linked to the hill height H.
5 THE SAN MARTINO TUNNEL FEM MODEL
Since a link of the observed behavior with stress
state can be guessed, a further set of analyses were
5.1 Settlements time history
conducted to evaluate the K0 effect. The reference
model with a 45° steep slope was studied at different Before comparing the fem model results with those of
initial stress states: K0 ⫽ 1.18 (OCR ⫽ 3), K0 ⫽ 0.873 the case study, a consideration must be done concern-
(OCR ⫽ 2) and K0 ⫽ 0.562 (NC). Besides two addi- ing the settlement time history.
tional analysis were run: the first at K0 ⫽ 1.18 for a At the time the monitoring started, the mass had
material having an higher friction angle (35°) thus already been disturbed by the excavation of the bomb
applying an OCR ⫽ 5 at the initial stage. The second shelter in the first 50 m. The monitoring cannot account
one was again the case K0 ⫽ 1.18 (OCR ⫽ 3), but a for the deformation phenomena already developed so
supplementary slope dipping 22° along the xy plane the analysis results have to be processed before com-
was introduced in the model. paring prediction and measured data.

303
6 CONCLUSIONS

A simplified tree dimensional numerical model was


realized to investigate the results of monitoring from
the excavation of S. Martino Tunnel in Ancona (Italy).
Morphological effects have been recognized playing
an important role in the stress field reconstruction.
A sensitivity analysis, based on the numerical model
developed, allowed to identify an influence zone induced
by the presence of a steep slope over the tunnel portal
which extension is proportional to the eight of the slope
itself. The study demonstrates that longitudinal settle-
ment curve can be predicted by adopting the profiles
45° slope K0 = 1.18 proposed in literature for cantilever excavation.
45° slope K0 = 0.873
45° slope K0 = 1.18 fric 35°
REFERENCES
45° +22° lat slope K0 = 1.18
Addembroke, T. I., Potts, D. M. & Putzrin, A. M. 1997. The
influence of pre-failure soil stiffness on the numerical
analysis of tunnel construction. Geotechnique, 47, No. 3:
197–204.
Figure 12. Longitudinal normalized excess settlement pro- Vermeer, P. A. & Brinkgreve, R. B. J. 1998. Finite Element
files from FEM analysis compared with Hsieh and Ou profiles Code for Soil and Rock Analysis. PLAXIS 7.0 and PLAXIS
(1998) for spandrel type excavations. 3D Tunnel manual. Balkema.
Franzius, J. N., Potts, D. M. & Burland, J. B. 2005. The influ-
ence of soil anisotropy and K0 ground surface move-
Distance from the edge of the slope: z/H ments resulting from tunnel excavation. Geotechnique, 55,
0 1 2 3 4 No. 3: 189–199.
-0,9
Gunn, M. J. 1993. The prediction of surface settlements pro-
Normalized settlements:

-1,0 files due to tunneling (eds. G. T. Houlsby and


-1,1 A. N. Schofield). Predictive soil mechanics: Proc. of the
Sv/Sv2D

Wroth Memorial symp., London: Thomas Telford, London.


-1,2
Hsieh, P. -G. and Ou, C. -Y. 1998. Shape of Ground Surface
-1,3 Settlement Profiles Caused by Excavation. Canadian
-1,4 Geotechnical Journal, 35, No. 6: 1004–1017.
-1,5
Katzenbach, R. & Breth, H. 1981. Nonlinear 3D analysis for
NATM in Frankfurt Clay. Proc. 10th Int. Conf. Soil.
-1,6 Mech. Found. Engng, Stockholm 1: 315–318.
CL section FG (ortho) CL section HG (ortho) Mair, R. J., Taylor, R. N. & Bracegirdle, A. 1993. Subsurface
CL section IG (ortho) section 1D from CL (ortho) settlement profiles above tunnels in clays. Geotechnique,
FEM analysis (reset after 2nd adv step) CL section FG (long)
43, No. 2: 315–320.
CL section HG (long) CL section IG (long)
section 1D from CL (long)
Panet, M. & Guenot, A. 1982. Analysis of convergence
behind the face of a tunnel. Tunnelling 82, The Institution
of Mining and Metallurgy, London: 197–204.
Figure 13. Longitudinal normalized calculated settlement Peck, R.B. 1969. Deep excavations and tunnelling in soft
profiles reset after two advancing steps compared with mon- ground. Proc. of the 7th Int. Conference on Soil Mechanics
itoring data. and Foundation Engineering. State of the art: 225–290.
Sociedad Mexicana de Suelos A.C.
5.2 FEM analysis results Segato, D. 2005. Influenza della topografia sulla evoluzione
dello stato deformativo del terreno nello scavo di gallerie
The existing excavation section being comparable with a bassa copertura. PhD Thesis. Università Politecnica delle
that of the new tunnel, the FEM analysis settlements Marche.
were reset after the first two advancement steps. Scarpelli, G. 1995. Physical processes and related instabilities
Due to the non-orthogonal arrangement between in overconsolidated clayey slopes. Proc. 11th European
Conference on Soil Mechanics and Foundation Engineer-
tunnel axis and slope edge a better matching between ing. Copenhagen.
computed and measured settlements can be achieved Simpson, B., Atkinson, J. H. & Jovicic, V. 1996. The influence
if distance orthogonal to the slope edge is taken of anisotropy on calculations of ground settlements above
into account instead of distance along the tunnel axis tunnels. Proc. Int. Symp. Geotechnical Aspects of Under-
(fig. 13). ground, Construction in Soft ground, London: 591–594.

304

You might also like