PREDICTED AND OBSERVED GROUND DEFORMATIONS DUE TO TBM

TUNNEL EXCAVATIONS ON THE IZMIR METRO PROJECT (STAGE 1)

Ahmet Unlutepe1,*, Volkan Tellioglu2,*, Basar Arioglu3

1
Bogazici University, KOERI, Geodesy Department, Istanbul, 34680, Turkey
[email protected]
2
Tunnel Engineer, Parsons Brinckerhoff, 69 Ann Street, Brisbane, 4001, Australia
[email protected]
3
Managing Director, Yapi Merkezi Construction Co., Camlica, 34676, Istanbul, Turkey
*
formerly with Yapi Merkezi Construction Co., Istanbul, Turkey

Keywords: Soft ground, EPB (earth pressure balance) TBM tunnel, Surface settlements

ABSTRACT
Stage 1 metro construction for the city of Izmir, which has the third highest population in Turkey,
has been successfully completed. It included a 2.75 km tunnel excavated by tunnel-boring machine
(TBM) using the earth pressure balance (EPB) method in soft ground conditions with shallow
overburden. This paper revisits the surface settlement values predicted by the empirical approach,
adds numerical analysis, describes the geotechnical instrumentation for field performance and then
compares the observed deformations with empirically and numerically predicted ground
deformations.

INTRODUCTION
With a population of 3.5 million, Izmir is the third biggest metropolis and one of the most
historically significant cities in Turkey. The total length of planned rail transport in the Izmir Bay
area is 45 km, according to the transportation master plan. The Yapi Merkezi – AdTranz
Consortium has successfully completed Stage 1 of metro construction for Izmir Metropolitan under
a design-and-build turnkey contract. With a total length of 11.5 km, the alignment’s peak capacity
will be 45,000 passengers per hour per direction in 2010. Total contract value was US$600m.
The project includes 2.75 km of tunnels excavated by tunnel-boring machine (TBM) using the earth
pressure balance (EPB) method. Yapi Merkezi subcontracted construction of the TBM tunnels to
YAPTAG JV (Yapi Merkezi – Wayss & Freytag AG Joint Venture). Of the 10 stations on the
project route, four are underground, and three are connected by TBM twin tunnels. The TBM tunnel
alignment is situated in soft ground with shallow overburden, and crosses populated and historic
built environment. The bore diameter of the completed TBM tunnels is 6.56 m.
This paper compares the tunnelling-induced, observed ground deformations with values predicted
by empirical methods and numerical analysis. It also describes the geotechnical instrumentation
used for field performance and, finally, compares the observed deformations with empirically and
numerically predicted deformations.

PROJECT ALIGNMENT AND BUILT-ENVIRONMENT

The 11.5-km-long project alignment was divided into its major elements: 1.7 km of mined tunnels,
1.4 km of TBM twin tunnels, 1.1 km of cut-and-cover tunnels, and the remainder was surface
works. Konak, Cankaya and Basmane stations were constructed using the cut-and-cover method.
Each station is approximately 200 m long, with excavation depths of 17 m, 20 m and 16 m
respectively. These underground stations were connected by EPB-type TBM twin tunnels. Four

1
tunnels were bored using a single EPB TBM that was launched from Basmane and Konak, then
received at Cankaya (Fig.1).
Civilisation dates back to 3000 BC in the Izmir area, and the majority of the TBM tunnel route had
to be constructed underneath this populated urban and historic environment in soft ground
conditions — the most challenging section of the whole project. Konak Station is close to a historic
tower clock, and one of the tunnels was excavated 10 m from this structure with 8 m to 9 m of soil
cover. Basmane Station was constructed next to the historic State Railways Station.
In addition, along a section of the alignment from Basmane towards Cankaya, twin tunnels had to
be excavated under a street lined on both sides with old masonry buildings of two- to three-storey,
as well as six- to nine-storey-high relatively new buildings with reinforced-concrete frames; some
of the buildings also had basements.

Figure 1: Project alignment (EPB twin tunnels are Figure 2: Alignment and typical cross-section for twin tunnels
situated between Konak and Basmane stations)

TUNNEL CHARACTERISTICS
TBM tunnel construction formed 12% of the total alignment in each direction. Four single tunnels
were excavated by one Herrenknecht EPB type of single-shield TBM that was launched from
Basmane and Konak, then received at Cankaya.
Overburden depth above the tunnel crown varies from 8 m to 13.5 m, which is approximately 1.3 to
2 times the bore diameter. The clearance between the two side-by-side twin tunnels varies between
5.4 m and 69 m approximately (Fig.2). Tunnel drive 3 and drive 4 are separated by a large distance
because it was proposed that an 18-m-deep underground pump house be constructed by another
contractor between the two tunnels.
A total of 2,292 rings were installed during TBM tunnel operations for four drives. The TBM
machine is 74.30 m long and weighs 550 tonnes. Total installed power is 1280 kW with 5000 kNm
maximum cutterhead torque. Table 1 shows the alignment characteristics and advance rates. The
maximum advance rate was achieved in Drive 4 - 25 rings per day, which equates to 30 m per day
(Fig.3). After each drive, the advance rate was observed to increase. This happened for three major
reasons. Firstly, the shield crew advanced along a learning curve in TBM operation, and gained
experience in the site-specific soil conditions. Secondly, the grout mix design was revised to get the
optimum performance. Thirdly, soil conditioning usage was optimised, improving the advance
rates.

2
 Table 1: The alignment characteristics and advance rates

Drive no. Route Total length Number of rings Average advance rate Minimum curve Maximum
[m] installed [m/day] radius [m] gradient [%]

Drive 1 Basmane-Cankaya 589.20 491 8.10 750.00 3.4

Drive 2 Basmane-Cankaya 591.60 493 12.56 751.75 4.0
Drive 3 Konak-Cankaya 813.60 678 13.10 249.86 2.2
Drive 4 Konak-Cankaya 756.00 630 19.20 298.15 4.0

Total 2750.40 2292

TBM advance rate

30

25
Advance rates [m/day]

20

15

10

0
0

50

100

150

200

250

300

350

400

450

500
Days

Drive 1 Drive 2 Drive 3 Drive 4

Figure 3: TBM advance rate for all four drives

The bore diameter is 6560mm, external shield diameter is 6520mm and finished lining intrados is
5720mm. The width of the lining ring is 1.2 m. Lining thickness is 300mm, and concrete class is
C45 (fck = 45MPa). All segments are conventionally reinforced. Each ring comprises 6 segments
and a key. Table 2 shows the details of geometric characteristics of the EPB shield and segmental
lining.
Table 2: Geometric characteristics of the EPB shield and segmental lining

Unit
Full-face excavation area [m2] 33.80
Length of shield [m] 9
Bore diameter [mm] 6560
Shield external diameter (at cutterhead) [mm] 6520
Conicity (total in diameter) [mm] 20
Thickness of shield [mm] 40
Gap between shield intrados and lining extrados [mm] 50
Lining extrados [mm] 6320
Lining intrados [mm] 5720
Segment thickness [mm] 300
Segment width [m] 1.20

GROUND CONDITIONS
The project area has complex topography and geology that consists of various soft ground
formations, from gravelly sands, sandy silts, and silty clays to very soft unconsolidated clays with
no boulders or bedrock formation. There are also sandy gravel beds with high artesian potential.
Water content of clay units, particularly between Konak and Cankaya, is nearly at the liquid limit.
Soils at Cankaya and Basmane are dominated by gravelly sands, sandy silts and silty clays. Konak
and Cankaya are close to land reclaimed over many centuries from the sea with a deep artificial fill
up to 6 m. Figure 4 shows the geotechnical profile of the alignment for the TBM tunnels.

3
 Figure 4: Geotechnical profile

There is a heterogeneous man-made fill layer varying from 2 m to 6m below the ground surface in
all subsurface investigation borings. Marine sediments are encountered at depths between 6 m and
15 m. These deposits comprise alternating layers of medium dense gravel, silty sand and dark gray
silty clay of medium consistency. Neogene aged gravelly clays and gravel bands are encountered
below 15 m under the surface. SPT blow counts of the clay layers are in the range of N30 = 14-30,
and increases with depth. There are water-bearing layers of sand and gravel with artesian pressure
within this clay layer. The bedrock is Miocene-age andesite. The unweathered bedrock is estimated
to be at a depths between 50 m and 60 m [Saglamer, A., 2005]. Silty soils have slightly more
plasticity, increased clay and reduced water content towards Basmane.
The TBM tunnels are located close to the coast of Izmir Bay. Sea level is 1 m to 3 m below the
ground surface, and groundwater level is almost at sea level. Tables 3a and 3b show the engineering
parameters for the stretch between Km 1+950 and Km 3+500 [Yapi Merkezi, 1996].
Table 3a: Engineering parameters for soils between Konak and Cankaya stations (Km 1+950 to Km 2+710)
Soil Layer γ Фu cu Ф' c' Es Ko k µ
[kN/m3] [º] [kN/m2] [º] [kN/m2] [MPa] [-] [m/s] [-]

F 19 - - 28.5 2 8-10 0.6 - 0.33

M 17 0 25 20 0 4.5-6 0.65 5x10-6 0.35
C 21 0 60 22 0 9 0.60 1x10-7 0.30
S 21 - - 37.5 - 40-60 0.40 2x10-5 0.30

Table 3b: Engineering parameters for soils between Cankaya and Basmane stations (Km 2+910 to Km 3+500)
Soil Layer γ Фu cu Ф' c' Es Ko k µ
[kN/m3] [º] [kN/m2] [º] [kN/m2] [MPa] [-] [m/s] [-]

F 19 - - 28.5 2 8-10 0.6 - 0.33

M 17 0 30 20 0 7-9 0.65 5x10-6 0.40
C 21 0 80 22 0 10-15 0.65 1x10-7 0.30
S-G 21 - - 37.5 - 60-80 0.40 2x10-5 0.20

PREDICTED GROUND DEFORMATIONS

Empirical methods
An empirical approach has taken to determining the subsidence that might occur over the tunnel
openings, and calculated values have been compared with the observed values. It is generally
accepted that the surface settlement profile above a tunnel in soft ground can be presented by a
Gaussian distribution curve. Gaussian curve provides a good approximation to the settlement trough
(Figure 5 modified from Arioglu et al, (2001) for twin tunnels). The maximum settlement over the

4
tunnel centreline, the equation for the settlement trough is given by Martos (1958), Schmidt (1969),
Peck (1969), and New & O’Reilly (1982):
(1) S = Smax exp(- x2 / 2 i2 )

where Smax : the maximum predicted surface settlement over the tunnel centreline, i : point of
inflection of the curve,
x : horizontal distance from centerline

The definition of maximum settlement and the shape of the subsidence trough can be determined
using the theory of volume loss (VL) combined with semi-empirical data based partially on
theoretical exceptions. The approach introduces the basic parameters of ground conditions, tunnel
geometry and TBM tunnel application into the analysis using coefficient for determination of
inflection point (i), percentage of volume loss (VL) and soil stability ratio (N) in relation to TBM
face support pressure and additional loads over the tunnel.
Volume loss (VL) is calculated as the ratio of the volume of observed settlement trough to the
volume of theoretical tunnel lining. Maximum settlement resulting from tunnelling can be
expressed with the following definition by New & O’Reilly (1991), Arioglu et al.(2002), and
Chapman et al (2004):

(2) VL = Vs / Vt = (2.5 i Smax) / (π/4 . D2) ⇒ Smax = (0.313 VL D2 ) / i

where VL: volume loss parameter (%), Vs: volume of surface settlement trough, Vt :
volume occupied by the tunnel, D: tunnel diameter.

Following the empirical approaches derived from the studies, case histories are used to calculate
point of inflection and stability ratio to determine the value of the lost volume:
(3) 2i/D = 1.392 (Zo / D) 0.704 Yapi Merkezi (1992), observed data from 22 cases

(4) i = 0.43 Zo + 1.1

i = 0.28 Zo – 0.1 O’Reilly & New (1982), for cohesive and non-cohesive soils

(5) i = 0.5 Zo O’Reilly & New (1982), Leca (1989), Mair and Taylor (1997), Chapman
et al. (2007)

(6) N=(Pe + γ Zo – Pface) / Cu Brooms & Bennermark (1967), Peck (1969), Arioglu et al.(1993)

(7) VL = A NB Leca (1989), and Arioglu et al.(1992), observed data from 70 cases
where A = 0.343 and B = 1.81
Other empirical equations based on field data used in estimation of maximum settlement are as
follows:
(8) Smax ≈ 0.0125 K (D2 / i) Schmidt (1969)

(8a) K = 0.87 exp(0.26 N) Yapi Merkezi (1992), observed from 30 cases

(9) Smax ≈ 0.950 N (D2 / β i) Arioglu et al. (1992)

(10) Smax ≈ 0.0125 VL (D2 / i) Schmidt (1969)

where Pe: additional pressure, γ: unit weight of the ground, Zo: depth of tunnel springline, D:
excavation diameter, N: stability number, Pface: face-support pressure, cu: cohesion, β: constant for
clay.

5
 Figure 5: Settlement trough for twin tunnels [Unlutepe, 2002]

The formulae (3), (4) and (5) above have been used to calculate the (i) values. The formulae (2),(8),
(9) and (10) have been used to calculate the predicted maximum settlement values for 25 sections
on the TBM route by using the volume loss and soil ratio method. The average of the calculated
values has been accepted as the (i) and (Smax) values for the further steps. Back analysis of the site
data determined that approximately 40% to 41% additional settlement occurred due to excavation of
the second tunnel. The average of the calculated settlement values has been adjusted accordingly,
and the results are given in Table 5.

Numerical analysis
A simplified two-dimensional (2D) finite-difference (FD) analysis has been used in a plane
perpendicular to the tunnel axis. A continuum model has been built using Itasca’s FLAC v5.0
software. A Mohr-Coulomb plasticity constitutive model and total stress analysis have been used in
the numerical simulation. The undrained behaviour of the cohesive soils has been considered.
Following an evaluation of the overburden, surface loads, distance between the twin tunnels,
geotechnical parameters and groundwater level, 10 cross-sections have been selected for numerical
modelling. The details of the analysis sections are given in Table 5. For the section where side-by-
side twin tunnels are spaced at more than three times their diameter, a single tunnel has been
modelled and analysed because no interaction is expected.
In the numerical model, four stages have been simulated in parallel to the ITA guidelines [ITA,
2000]: i) initialisation of the in situ stresses; ii) excavation supported by shield; iii) tail void
grouting; and iv) installation of permanent lining.
The majority of the tunnel route was excavated in silts and soft clays, particularly between Konak
and Cankaya where the water content was nearly at the liquid limit. The expected failure zone
would extend ahead of the tunnel face which makes immediate face support necessary (Fig.6). A
longitudinal displacement profile calculation has not been coupled to the analyses because the
tunnels are situated in soft soil with shallow overburden, and the convergence–confinement method
is a solution for deep tunnels that does not account for a free surface located near the crown. No
stress relaxation has been taken into account in the numerical calculations, and it has been assumed
that TBM would provide the ideally full face-support pressure.

6
 Figure 6: Expected failure mechanism for Figure 7: Basic dimensions of the soil-shield-lining
unsupported tunnel along the alignment in silty soils interface (segments not to scale)

The shield is assumed as rigid body, so a fictitiously high stiffness has been used in the simulation.
The soil-shield interface has been modelled with 20mm overcut gap (in radius) of shield. Conicity
of the shield has also been modelled as 20mm in diameter. Total tail void is 120mm in radius
(Fig.7).
Grouting pressure has been taken as a total of hydrostatic pressure increased by 1 bar minimum
typically, plus the surface loads, if any. The pressure has been applied to the periphery of the
excavation area because of the varying intensity between the crown and invert of the tunnel. Tail
void grout and permanent segmental liner elements have been modelled as linearly elastic material.
For the void grout, short-term elasticity modulus Eg= 50MPa, poisson’s ratio υ= 0.3 and unit weight
γ= 21 kN/m3 have been used. For the lining, bending stiffness EI= 96.66 MN/m2, and axial stiffness
EA= 12888 MN have been considered.

OBSERVED GROUND DEFORMATIONS

Instrumentation and monitoring of field performance
A geotechnical monitoring program was implemented for all underground works on the project,
including TBM tunnels. A total of 25 monitoring sections (MS1 through MS25) were established
along the TBM tunnel alignment. Settlement bolts, precise levelling, inclinometers, tilting
measurements and crackmeters were used to monitor surface settlements, segmental lining
convergence, movement of structures and crack propagation. Historic structures within the
influence zone of tunnels, such as Clock Tower, Konak Mosque, Vakiflar building near Konak
Square, and State Railways Station building, were specifically instrumented. Tilting measurements
for the tower and minaret of Konak Mosque were taken, and crackmeters were installed onto
historic Vakiflar building and Basmane State Railways Building to observe tunnelling-induced
deformations. Table 4 summarises instrumentation developed for TBM tunnels.
Table 4: Instrumentation summary for TBM-induced settlement monitoring
Abbreviation Element of monitoring Number of points Remarks

Surface and building bolts and

MS-S-Inc 420
inclinometer
Surface and building bolts, precise level 112 At surface Located at the each corner of buildings and
MS-S
readings 315 On buildings closely spaced at the surface

MS-Inc Inclinometers 7 In the ground

MS-D Tilting observation by total station 2 On buildings For Clock Tower and Konak Mosque
Vakıflar Building, Basmane State Railways
MS-C Crack observation by crackmeters 2 On buildings
Building
MS-O Convergence measurements Inside the completed segmental rings

A total of 427 settlement bolts were installed to monitor surface settlements; 112 were set at the
surface on the axes of the instrumented sections, and 315 were placed on the corner columns of the
buildings. Measurements were made by a Zeiss Ni2 level instrument with a precise reading

7
accessory and two invar rods. A total of 5263 precise level readings were taken with respect to the
vertical network along the tunnels. The Sinco 50310900 model inclinometers were installed on the
ground surface at the sections MS6, MS8, MS9/A, MS10/A, MS11, MS12 and MS21, in order to
monitor lateral movements in the ground along the tunnels; 60 readings were taken during
construction. Convergence measurements were taken inside the tunnels by a Nedo mess-fix
telescopic (8 m) measuring ruler. The readings varied between 0 mm and 10 mm in vertical and
horizontal directions. Figure 8 shows the typical layout for instrumentation. An average of two
readings per running metre of tunnel construction were taken. The average number of
measurements per day was 20.

Figure 8: Typical instrumentation layout Figure 9: Measurement (reading) frequency for

[Unlutepe et al., 2008] geotechnical monitoring of TBM tunnels [Unlutepe, 2002]

where S: settlement bolt (on the ground), Sb: settlement bolt (on the buildings), I: inclinometer, Cr:
crackmeter, T: tiltmeter

Reading frequency
Figure 9 shows the measurement (or reading) frequency that was followed in the field monitoring of
TBM tunnels. The zero readings were taken once the installed instrument was installed and
stabilised. The initial readings were taken no later than when TBM arrived at a point 70 m from a
particular instrumented monitoring section (MS). The second measurement was taken when the
TBM was between 70 m and 50 m from the MS. Daily readings were taken while the TBM was
passing 50 m before and after the monitoring section. A single measurement was taken after the
TBM passed and was situated 50 m to 70 m from the monitoring section. The frequency of
subsequent measurements was decided by the team, depending on the magnitude and tendency of
ground deformations.

DISCUSSION OF RESULTS
The summary of predicted and observed surface settlements is given in Table 5. In general, the
predicted maximum settlements by empirical and numerical methods have yielded larger values
than the observed ones. The settlements obtained by numerical approximations are larger than the
empirical results. Despite the differences in magnitudes, both empirical and numerical methods
have shown similar tendency of change with the observed values along the alignment. All values of
predicted and observed settlements have concentrated between 5 mm and 30mm (Fig.10).
According to the average of six pairs of values in the numerical analysis, the additional settlement
caused by the second tunnel was predicted to be 35% of the total maximum settlement, which
means that approximately 65% of the total settlement was expected to occur once the first tunnel
was complete. The average of 11 observed values indicates that this ratio is actually 41% of the
total settlement.

8
 Predicted and observed maximum surface settlements along the alignment Maximum settlement readings along the alignment
0 5

-5 0

-10 -5

Settlement [mm]
Settlement [mm]

-15 -10

-20 -15

-25 -20

-30 -25

-35 -30

1+960

2+000

2+037

2+070

2+122

2+153

2+208

2+245

2+280

2+309

2+408

2+410

2+463

2+540

2+640

2+700

2+915

2+954

3+089

3+202

3+311

3+410

3+453

3+486
1+960

2+000

2+037

2+070

2+122

2+153

2+208

2+245

2+280

2+309

2+408

2+410

2+463

2+495

2+540

2+640

2+700

2+745

2+870

2+915

2+954

3+089

3+202

3+311

3+410

3+453

3+486
Chainage Chainage
Empirical - adjusted Numerical - both tunnels complete Observed - both tunnels complete At the ground surface On the buildings

Figure 10: Predicted and observed maximum Figure 11: Readings obtained from settlement bolts installed
surface settlements along the alignment at the ground surface and on the buildings
[Unlutepe et al., 2008]

Table 5: Summary of predicted and observed surface settlements

Approx. depth
Average Hor. spacing Min. ground Hor. spacing Cover to
Monitoring Direction of groundwater
Chainage Drive no depth to between tunnel cover above to bore dia bore dia Max. transverse surface settlement [mm] Volume loss
section of Drive table from
tunnel center centerlines crown ratio ratio
surface

Zo a C a/D C/D Predicted- empirical Predicted- numerical Observed Empirical Observed

Both Both
1st tunnel 1st tunnel
[m] [m] [m] [m] Average Adjusted tunnels tunnels [%]
complete complete
complete complete

1+950 KONAK UNDERGROUND STATION

MS-1 1+960 3 and 4 11.42 11.73 8.14 1.79 1.24 2 -8 -11 -8 -12 0.51 0.55
MS-2 2+000 3 and 4 11.89 11.73 8.61 1.79 1.31 2 -10 -13 -10 -21 0.62 1.00
MS-3 2+037 3 and 4 12.17 11.72 8.89 1.79 1.35 2 -11 -15 -14 -24 -9 -19 0.73 0.92
MS-4 2+070 3 and 4 12.69 11.72 9.41 1.79 1.43 2 -12 -18 -14 -26 -9 -20 0.90 1.00
MS-5 2+122 3 and 4 13.05 12.12 9.77 1.85 1.49 2 -14 -20 -5 -16 1.03 0.82
from Konak to Cankaya

MS-6 2+153 3 and 4 13.67 13.71 10.39 2.09 1.58 2 -17 -24 -18 -28 -9 -10 1.28 0.53
MS-7 2+208 3 and 4 13.98 22.32 10.70 3.40 1.63 2 -18 -25 -5 -7 1.36 0.38
MS-8 2+245 3 and 4 14.25 33.08 10.97 5.04 1.67 2 -19 -19 -14 -14 -5 1.05 0.28
MS-9/A 2+280 3 13.88 46.10 10.60 7.03 1.62 2 -18 -18 -8 0.97 0.43
MS-9 2+309 4 14.95 62.21 11.67 9.48 1.78 2 -23 -23 -28 -28 -5 1.33 0.29
MS-10/A 2+408 3 15.20 75.12 11.92 11.45 1.82 3 -24 -24 -26 1.40 1.52
MS-10 2+410 4 16.15 75.45 12.87 11.50 1.96 3 -29 -29 -13 1.78 0.80
MS-11 2+463 4 16.53 67.93 13.25 10.36 2.02 3 -31 -31 -23 -23 -11 1.94 0.69
MS-11/A 2+495 3 16.78 71.03 13.50 10.83 2.06 3 -32 -32 -21 2.04 1.34
MS-12 2+540 3 and 4 15.75 41.44 12.47 6.32 1.90 3 -27 -27 -30 -30 -7 1.63 0.42
MS-13 2+640 3 and 4 14.27 16.01 10.99 2.44 1.68 4 -19 -27 -2 -7 1.50 0.39
MS-14 2+700 3 and 4 13.02 11.93 9.74 1.82 1.48 4 -14 -20 -7 -7 1.03 0.36
2+745
CANKAYA UNDERGROUND STATION
2+870
MS-15 2+915 1 and 2 13.53 11.72 10.25 1.79 1.56 4 -13 -18 -3 -3 0.95 0.48
from Basmane to

MS-16 2+954 1 and 2 13.82 11.72 10.54 1.79 1.61 5 -14 -19 -21 -34 -12 -16 1.03 0.86
MS-17 3+089 1 and 2 13.62 11.72 10.34 1.79 1.58 5 -13 -18 -24 -29 -4 -7 0.96 0.37
Cankaya

MS-18 3+202 1 and 2 13.09 11.72 9.81 1.79 1.50 5 -11 -19 -4 -4 0.83 0.17
MS-19 3+311 1 and 2 12.83 11.73 9.55 1.79 1.46 5 -11 -18 -5 -5 0.77 0.21
MS-20 3+410 1 and 2 12.33 11.78 9.05 1.80 1.38 5 -9 -16 -15 -21 -8 -7 0.66 0.29
MS-21 3+453 1 and 2 12.13 11.78 8.85 1.80 1.35 5 -9 -15 -12 -14 0.61 0.57
MS-22 3+486 1 and 2 12.02 11.72 8.74 1.79 1.33 5 -9 -14 -8 -8 0.57 0.32
3+497 BASMANE UNDERGROUND STATION

Good agreement has been observed between the records obtained from the settlement bolts installed
at the ground surface and those from on the buildings (Fig.11).
Figure 12 shows the predicted and observed transverse settlements. As mentioned above, four
stages, based on the TBM operations, have been used in the numerical model. After each step,
additional settlements have been obtained. The simulation of the overcut gap has shown the full
closure of soft soil onto the shield, which causes a considerable percentage of the total surface
settlement. Averaging the results of 10 sections showed that approximately 36% of the total
settlement occurred along the shield, which is close to the current knowledge of distribution (40%
to 50%, [ITA, 2006]). The same set of outputs has indicated that an average value of approximately
60% of the total settlement has occurred in the tail void, which is higher than common practice
values (30% to 40%, [ITA, 2006]). In six analysed sections (MS3, MS4, MS6, MS16, MS17 and
MS21), the results of the models for twin tunnels located side-by-side , with a centre-to-centre
distance of 1.8 to 2 diameters, have shown that the settlement trough is not symmetrical about the

9
tunnel centreline and that there is a tendency for the point of maximum settlement to shift towards
the second tunnel excavated, particularly for the sections between Konak and Cankaya. This
behaviour can be interpreted as due to the interaction between closely spaced side-by-side twin
tunnels. The ground close to the first tunnel has been disturbed by the construction of the first
tunnel, reducing the stiffness of the ground in this area, and subsequent excavation in this pre-
strained soil results in larger settlements. This effect tends to decrease slightly in the sections
towards Basmane where the soil stiffness increases.

Predicted and observed transverse ground surface settlements Predicted and observed transverse ground surface settlements
Monitoring section MS6 (Ch: 2+153) Monitoring section MS9 (Ch: 2+309)
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30

0 0

-10 -10
Settlement [mm]

Settlement [mm]
-20 -20

-30 -30

-40 -40
1st 2nd

-50 -50
Distance [m] Distance [m]

Shield passing- 1st Tail void grouting- 1st Permanent lining- 1st Shield passing- 2nd
Tail void grouting- 2nd Permanent lining- 2nd Empirical- 1st complete Empirical- both complete Shield passing Tail void grouting Permanent lining Empirical- max Observed- max
Observed- 1st complete Observed- both complete 1st excavated 2nd excavated

Predicted and observed transverse ground surface settlements Predicted and observed transverse ground surface settlements
Monitoring section MS12 (Ch: 2+540) Monitoring section MS20 (Ch: 3+410)
-30 -20 -10 0 10 20 30 -30 -20 -10 0 10 20 30

0 0

-10 -10
Settlement [mm]
Settlement [mm]

-20 -20

-30 -30

-40 -40
1st 2nd
Note: Data not available for the
"Observed- 1st complete" curve
-50 -50

Distance [m] Distance [m]

Shield passing- 1st Tail void grouting- 1st Permanent lining- 1st Shield passing- 2nd
Shield passing Tail void grouting Permanent lining Empirical- max Observed- max Tail void grouting- 2nd Permanent lining- 2nd Empirical- 1st complete Empirical- both complete
Observed- both complete 1st excavated 2nd excavated

Figure 12: Predicted and observed transverse settlements

The results of the numerical model have showed wider settlement troughs between Konak and
Basmane due to the predominant cohesive behaviour, while the width of the troughs has slightly
reduced due to the intrusion of the sand bands towards Basmane; this agrees with the actual
behaviour.
In spite of the large clearance between the twin tunnels and relatively higher cover-to-depth (C/D)
ratios, large settlements have been observed between the monitoring sections MS10A and MS11A.
There was no abnormality reported by the shield crew while passing this location. A possible reason
is that there was continuous dewatering performed by another contractor due to the 18-m-deep
excavation for a pump house located in between the twin tunnels at around chainage 2+400, where
the maximum settlement of 26 mm was recorded. The depth of excavation was almost at the same
level of the tunnel invert. The geotechnical profile indicates sand bands and pockets underneath the
tunnels over this area that might create a path for quick drainage and an increase in the effective
stress, resulting in larger settlements than anticipated. In addition, this stretch of the alignment has
the minimum curve radius (R=250m). Therefore, additional volume loss due to the curved tunnel
alignment also caused an increase in the settlement. Another contributing factor was that major
underground utilities were located underneath the street near the twin tunnels in this area.
The predicted values for volume loss (VL) by empirical methods have been found to vary in a range
from 0.51% to 2.04%, with an average of 1.10%; however, according to the field monitoring results
along the alignment, the VL varied between 0.17% and 1.52%. The average of the observed VL is
0.60%. These values are in agreement with the recent practices with closed-face mechanised

10
tunnelling, i.e. VL is 0.5% for sands and gravels, and VL ranges between 1% to 2% in soft clays
[Guglielmetti et al., 2009].
There was no functional or structural damage caused by the construction of the twin tunnels. Of 150
buildings monitored in the zone of influence of both tunnels, 12 were registered as heritage. Only a
few hairline cracks were observed on the State Railways Building at Basmane and the Vakiflar
building at Konak. Five crackmeters on each building were monitored continuously, but no crack
propagation was recorded. The degree of severity ranged between negligible and very slight
[Burland et al., 1977].

The effect of TBM face-support pressure on the surface settlements

Face pressures applied for all four drives varied between 1.6 and 2.4 bars (160 to 240kPa) at the
tunnel springline. Figure 13 shows the variation of surface settlements (at tunnel centerline) with
respect to applied face pressure (at tunnel springline).

Figure 13: Surface settlement vs applied face-support pressure [Arioglu et al., 2002]

Soil conditioning
Foam and bentonite soil conditioning was used in EPB tunnelling to improve the interaction
between soil and groundwater, depending on the ground conditions. The first drive was in silty soils
and clay. The foam consumption started at an uneconomically high level of 1500 lt/m3; by using
bentonite slurry, consumption was reduced to 500 lt/m3, and later to 300 lt/m3. Torque of the
cutterhead was also reduced by this modified foam conditioning, which saved energy, minimised
tool wear and stabilised support pressure. This resulted in very low surface settlements.
Soil conditioning improved the efficiency of EPB tunnelling by making it possible to excavate,
support and transport the muck within economical boring parameters. It also induced a stable
support pressure in the working chamber during boring [Jancsecz et al., 1999]. Soil conditioning
consequently helped to minimise the surface settlements.

CONCLUSION
A total length of 2.75 km of twin EPB tunnels was successfully constructed in soft-ground
conditions with shallow overburden on the Izmir Metro Project (Stage 1). During construction,
9356 readings were taken from 676 points over 25 monitoring sections. The comparison of the
observed surface settlements with the predicted values by empirical methods and numerical analysis
has been presented. In general, the predicted maximum settlements have yielded larger values than
the observed ones. However, despite the differences in magnitudes, both empirical and numerical
data have shown similar tendencies of change from the observed values along the alignment.
The soil experiences different states of stress and strain throughout the TBM-driving process. Face
pressures, grouting pressures, consolidation of the soil around the tunnel face, conicity of the shield,
rate of advance, proximity of other deep foundations or structures are all important factors in TBM
tunnelling, and each of these factors has considerable impact on the prevalence of surface

11
settlements. Simplified 2D numerical models could not capture the complications generated by
these factors. The simplified 2D models are easy to use, faster to iterate and may be useful for
interpretation of common behaviour and deformations for concept and preliminary studies.
However, numerical results should always be run in parallel with them and checked against the
empirical methods. 3D (even 4D) advanced models look promising for realistic simulation of TBM
driving and installation steps, which may be used for the detailed design.
The low values of surface settlements have indicated a successful TBM operation — that is, face-
support pressure, tail void grouting pressure, soil conditioning and other factors have been properly
planned, adjusted and implemented, which consequently has delivered a safe tunnelling experience
for the built environment.

ACKNOWLEDGEMENTS
This paper reflects the view of the authors, who are solely responsible for the reliability and
accuracy of the information presented. The authors would like to express grateful
acknowledgements to the Founder and Honorary Chairman of Yapi Merkezi, Dr.Ersin Arioglu, who
encourages and supports technical and scientific research integrated with commercial operations.
We would like to thank Prof. Dr. Ergin Arioglu, the Director of Research and Development in Yapi
Merkezi for his academic contributions to the subject matter some of which were cited in this paper.
We also thank the Chairman of Yapi Merkezi, Mr. Emre Aykar for his support during and after the
project.

REFERENCES
Arioglu, E., Arioglu B., Arioglu E., Odbay O. (1992), Determination of Parameters of Surface Subsidence Caused by
Shallow Excavations, Turkish Congress on Soil Mechanics and Foundation Engineering, Istanbul.
Arioglu, E., Arioglu B., Arioglu E., Odbay O. (1993), Determination of Parameters of Surface Subsidence Caused by
Tunnel Construction by Means of Semi-analytical Approaches, XIII Mining Assembly, Turkey.
Arioglu, B., Yuksel, A., Arioglu, E. (2000), Excavation Parameters and Evaluations of Nene Hatun Tunnel-Izmir Metro
(in Turkish), Proceedings of the 5th National Rock Mechanics Symposium, Isparta , pp. 21-33.
Arioglu, E., Yilmaz O., (2001), Practical Mining Problems, TMMOB, Chamber of Mining Engineers, Ankara, Turkey.
Arioglu, B., Yuksel A., Kurtuldu S. (2002), Arioglu E., NATM, EPBM and Cut-and-Cover Tunneling Applications in
the Project of Izmir Metro, Felsbau 20, No:4 pp 14-22.
Broms, B.B., Bennemark, H. (1967), Stability of clay at vertical openings, ASCE, Journal of Soil Mechanics and
Foundation Engineering Division, SMI 93, 71–94.
Burland, JB., Broms, B. and De Mello, VFB. (1977), Behaviour of Foundations and Structures, Proceedings IX
ICSMFE, Tokyo.
Chapman, D.N., Rogers C.D.F., Hunt D.V.L. (2004), Predicting the Settlements Above Twin Tunnels Constructed in
Soft Ground, Proceedings of the ITA-AITES World Tunnelling Congress, Singapore, Tunnels and Underground Space
Technology 19, 378 (CD-ROM).
Chapman, D.N., Ahn S.K., Hunt D.V.L. (2007), Investigation Ground Movements Caused By the Construction of
Multiple Tunnels in Soft Ground Using Laboratory Model Tests, Canadian Geotechnical Journal Vol:44, No.6 June,
p:631-643.
Guglielmetti, V., Grasso, P., Mahtab, A., Xu, S. (2009), Mechanized Tunnelling in Urban Areas, Taylor & Francis
ITA (2000), Guidelines for the Design of Shield Tunnel Lining, France.
ITA (2006), Settlement Induced by Tunneling in Soft Ground.
Jancsecz S., Krause R., Langmaack L. (1999), Advantages of Soil Conditioning in Shield Tunneling: Experiences of
LRTS Izmir, ITA, Oslo.
New, B.M., O’Reilly, M.P, (1991), Tunnelling-induced Ground Movements: predicting their magnitude and effects, 4th
International Conference on Ground Movements and Structures, Cardiff, invited review paper, pp. 691–697.

12
 Leca, E. (1989), Analysis of NATM and shield tunneling in soft ground, Ph.D. Thesis, Virginia Polytechnic Institute and
State University, B1acksburg.
Martos, F.,(1958), Concerning an Approximate equation of the subsidence Trough and Its Time Factors.
In:International Strata Control Congress, Leipzig. PP.191-205
O’Reilly, M.P., New, B.M, (1982), Settlements above Tunnels in the United Kingdom: their magnitude and prediction,
Tunnelling’82, London. IMM, pp. 173–181.
Peck, R.B. (1969), Deep Excavations and Tunneling in Soft Ground, 7th International Conference on Soil Mechanics
and Foundation Engineering, Mexico City State-of-the-Art volume, pp. 225–290.
Saglamer, A. (2005), ITA/AITES –Training Course: Tunnel Engineering, Istanbul.
Schmidt, B. (1969), Settlements and Ground Movements Associated with Tunneling in Soils, Ph.D. Thesis, University of
Illinois, Urbana.
Unlutepe, A. (2002), Monitoring Vertical and Horizontal Displacements During the Construction of Izmir LRTS NATM
and EPBM Tunnels, Ph.D. Thesis, Bogazici University, Geodesy Department, Istanbul.
Unlutepe, A., Ozener, H. (2003), A Comparison of Predicted and Measured Values of Ground Settlements for the Izmir
Metro Konak-Basmane EPBM Tunnels, 9th Turkish National Mapping, Technical and Scientific Assembly, Ankara.
Unlutepe, A., Ozener, H. (2008), Geotechnical Measurements at Izmir LRT System Tunnels, Tunneling and
Underground Space Technology, Volume 23, Issue 6, Pages 734-741.
Yapi Merkezi (1992), Prediction of Ground movements during the Construction of Istanbul Metro, Internal Report Yapi
Merkezi, Istanbul.
Yapi Merkezi (1996), Izmir LRTS — Geotechnical Report (Km 1+550 to Km 3+800), Report No.9210-TR-J058.T,
Istanbul.

13
All in-text references underlined in blue are linked to publications on ResearchGate, letting you access and read them immediately.