Numerical Analyses of Sequential Tunneling in Chicago

Glacial Clays

Abdolreza Osouli1, Siavash Zamiran2, Sree Kalyani Lakkaraju3

1PhD, P.E., Assistant Professor, Dept. of Civil Engineering, Southern Illinois University Edwardsville,

Edwardsville, Illinois, USA, Email: aosouli@siue.edu, Phone: +1 (618) 650-2816 2PhD Candidate, Instructor, Dept. of Civil Engineering, Southern Illinois University Carbondale,

Carbondale, Illinois, USA, Email: zamiran@siu.edu, zamirans@gmail.com, Website: www.zamiran.net,

Phone: +1 (618) 334-4572 3Graduate Student, Southern Illinois University Edwardsville, Edwardsville, Illinois, USA

48th US Rock Mechanics / Geomechanics Symposium held in Minneapolis, MN, USA, 1-4 June 2014

Postprint version

Citation: Osouli, A., Zamiran, S., & Lakkaraju, S. K. (2014). Numerical Analyses of Sequential

Tunneling in Chicago Glacial Clays. In 48th U.S. Rock Mechanics/Geomechanics Symposium.

Minneapolis, Minnesota: American Rock Mechanics Association.

ABSTRACT: The key factor in successful construction of urban tunnel projects is selecting a suitable

excavation procedure in soft ground. The choice of the excavation procedure strongly influences the cost and time for

tunnel construction. The aim of this paper is to analyze delay in liner installation for tunnel construction using one-

and two-pass lining system via three-dimensional finite element numerical modeling. The tunnel is assumed to have

a circular cross section in Chicago glacial clays with a diameter of about 3.8 m and a centerline at a depth of 10.5 m

below the ground surface. The soil profile consists of compressible clay deposits (i.e. Blodgett and Deerfield) and a

relatively incompressible hard silty clay stratum (i.e. Park Ridge) and the tunnel alignment is assumed within Deerfield

compressible clay layer. The thickness of the inner liner is assumed 12.7 cm. The constitutive model used to

characterize the clays in the simulation is the Modified Cam Clay model. The soil stratigraphy was assumed to be

uniform within each layer. The one and two-pass lining systems are modeled in ABAQUS using Model Change option

and Load Reduction method, respectively. For load reduction method, the concentrated loads in equilibrium with the

initial stress field are applied along the perimeter of the tunnel. These forces were sequentially reduced after initial

liner placement to evaluate the creep effect. The results of 3D finite element analyses with emphasis on ground

stability, axial and radial deformations of the tunnel, and stresses transferred to the liner are presented for both

analyses.

Keywords: One-pass and two-pass lining systems, Chicago Glacial Clays, Load Reduction Method, Modified Cam

Clay material model

1. INTRODUCTION

Tunnels are constructed with excavating soil or rock material and providing lining support. Tunnels are

excavated in various projects such as highways and sanitary system construction, or subways. Tunneling in

urban areas is challenging because in addition to presence of plenty of common uncertainties such as soft

soil, sensitiveness of ground deformation, environmental difficulties, etc, the foundation of surrounding

structures and effect of ground deformation on adjacent structures need to be considered.

The tunneling construction methods could be divided into two distinct categories including one-pass and

two-pass systems. In two-pass tunneling method the steps of excavation and installation of lining are

conducted separately in two distinct phases. At the first phase, the excavation to a certain length would be

done and temporary lining is installed. In the next phase, the lining installation is conducted. Although the

temporary lining systems could be varied in different projects, the permanent lining system which is used

in this method is mostly cast-in-place concrete lining. This method has different advantages including the

capability of constructing tunnels in different plan sections, proportionally less expensive tunneling and

lining instruments and potential of refining the non-uniformity of excavation process. However, the method

has some disadvantages considering erosion of the lining in adjacent of water and corrosion materials,

concrete casting problems after reinforcement installation and the inherent time consuming situation of the

two-pass construction procedure [1].

The alternative tunneling method which is used in a wide range worldwide is one-pass method which is

defined as coincidental execution of excavation and lining in one phase with the use of tunneling boring

machine (TBM). In this method precast concrete liner is utilized for tunnel supporting system which could

reduce the cracking and chemical attack of the medium and increase the time efficiency of the progress.

However, expensiveness of high-tech tunneling tools and the necessity of high level of professionalism are

inevitable in this method.

A reliable prediction of tunneling behavior prior to the start of the project could considerably reduce the

unexpected risks and enhance project safety. Numerical modeling using finite element method is one of the

applicable tools that was utilized to predict different tunneling projects in the recent decades. Negro and

Queiroz (2000) indicated that in 96 percent of tunneling studies finite element method techniques were

implemented [2]. For example, Farias et al. (2004) investigated the effects of New Austrian Tunneling

Method (NATM) on ground settlement in a three dimensional simulation [3]. NATM method is assumed

as the two-pass tunneling method because of a two separate phases of excavation and lining installation.

The authors conducted three-dimensional finite element analysis on a sample tunnel and they concluded

that the settlement of the tunnel crown decreases in the adjacent of the tunnel face even if a permanent

lining system is not installed. Karakus and Fowell (2005) investigated the tunneling process of the

Heathrow Express Trial Tunnel, which was also conducted based on one passing tunneling system in

London Clay [4]. The results were compared to experimental values extracted by the instrumentation

equipments in Heathrow Trial Tunnel. The results indicated a reasonable agreement between the values of

numerical modeling and experimental data. Svoboda and Bohac (2010) used the hypoplastic model, which

is an extension of Modified Cam-Clay model, for clays in simulating of Královo Pole tunnels located in

Czech Republic [5]. The excavation method was based on NATM and the stratum contained stiff clay

material. Wang et al. (2012) conducted a survey on the short-term and long-term response of a shallow

tunnel in clayey soil [6]. The investigators used finite element method for numerical analysis of the model

and compared the numerical results with the field observed data. Furthermore, creep behavior of clay was

studied for long-term function of the tunnel. The authors indicated thatthe settlement of the tunnel should

be evaluated based on creep and consolidation of the medium in parallel with elasto-plastic settlement of

the system.

In parallel with technical studies on two passing tunneling system, there is plenty of investigations which

are focused on one passing tunneling technology. One of the primary numerical investigations on the one

pass method was conducted by Swoboda and Abu-Krisha (1999) [7]. The authors conducted finite element

analysis of tunneling based on one pass method by tunneling boring machine in three-dimensional analysis.

They focused on the excess pore pressure generation around the tunneling sequence. Their results indicated

that the slurry and grouting pressure of the tunneling procedure changes the pore pressure distribution

around the excavated zones considerably. Lambrughi et al. (2012) developed a numerical model in a case

study of Madrid metro extension project from 1995 to 2003 which was based on one-pass system. The

authors investigated the tunnel stability using constitutive models such as Linear-Elastic, Mohr-Coulomb

and Modified Cam-Clay [8]. They compared the results with instrumentation field measurements and

indicated that Modified Cam-Clay model could predict the deformational behavior of the medium better

than Mohr-Coulomb and Linear-Elastic models.

In this investigation, the effect of the two tunneling methods in Chicago glacial clays are studied using

numerical modeling including one-pass and two-pass methods. In the one-pass lining practice, the

construction time would be less and the pre-cast concrete or cast-iron rings are being used in a single phase

as the excavation progresses. In the two-pass lining method, initial and secondary lining systems are used.

The initial lining usually consists of fiber reinforced or plain shotcrete and the secondary lining consist of

either shotcrete or cast-in-place reinforced concrete. The installed initial lining is not often considered a

component of the long-term load-carrying lining system. The finite element analyses of the modeling were

done using commercial program ABAQUS version CAE 6.11.

2. NUMERICAL MODELING

The tunnel is assumed to have a circular cross section characterized by an excavation diameter of 3.8 m and

situated at a depth of 10.5 m below the ground surface. The dimensions of the model are 30 m in length, 21

m in height and 21 m in width to eliminate any boundary condition effect on the model. The thickness of

the inner liner is assumed at 12.7 cm. Only five meters excavation was simulated to demonstrate the

tunneling behavior comparison of one-pass and two-pass methods. The finite element meshing of the

numerical model is presented in Fig. 1. In this figure, three vectors illustrate the sections, which were

monitored during analyses.

Fig. 1. Finite element meshing of the tunnel

The material properties are derived from the experimental research results of Finno and Calvello (2005)

and Finno and Chung (1992) [9, 10]. The soil profile consists of three main clayey soils named Blodgett,

Deerfield and Park Ridge layers from top to bottom. The Park Ridge stratum is relatively stiff clay layer.

In this investigation, the Cam–Clay constitutive model was used for simulation of the layered clays. Table

1 presents the material properties of the strata based on Cam-Clay constitutive model.

Ground surface

z

yx

Table 1. Modified Cam-Clay properties for Chicago clays

Parameters Layer 1 Layer 2 Layer 3

Initial void ratio 1.55 0.8 0.68

Lateral earth pressure

coefficient

0.54 0.54 0.46

Specific weight (kN/m3) 1600 1320 1398

Poisson's ratio 0.25 0.25 0.2

Coefficient of Permeability

(m/s)

3.1E-10 3.1E-10 3.1E-10

к coefficient in Cam Clay

model

0.0826 0.0449 0.0304

Normal compression line

slope

0.143 0.076 0.050

Stress ratio M 0.984 1.243 1.322

Wet yield surface size β 1 1 1

Flow stress rate K 1 1 1

Initial yield surface size (kPa) 65.2 92.3 167.0

The one-pass lining system simulation was analyzed using model change option of ABAQUS. The

simulation of tunnel construction is processed in five one meter sequences. These five sections are labeled

and shown in Fig. 2 .In every phase of simulation, the excavation of 1 m soil and liner installation is modeled

to be completed in one day. A time lag of one day is considered to represent typical construction practice

and monitor soil movement. The simulation of tunnel construction is continued by repeating the sequence

for rest of the 1 m sections.

Fig. 2. Cross section of the excavation sequences

Excavation direction

Dia

met

er 3

.80 m

Y=-5 Y=0

1m 1m 1m 1m 1m

Seq

uen

ce 1

Seq

uen

ce 2

Seq

uen

ce 3

Seq

uen

ce 4

Seq

uen

ce 5

In first step of this analysis, the liner elements are deactivated and dead load is applied to the rock mass

representing the geostatic condition. The geostatic step definition is used to specify the initial step. In Phase

1 of tunneling, the soil mass is deactivated for first 1 m section in the first step and after a day the lining

layer is installed permanently in the third step. The last step in phase 1 is step 4, where the model is analyzed

for ground deformation for a time period of 1 day. The simulation of tunnel construction is continued by

repeating all above mentioned steps for other sequences of tunneling. Once the tunnel construction phases

are modeled, the ground deformations are monitored for durations of ten days, one month and six months.

Consequently, each model consists of a total of 19 steps.

The two-pass lining system was simulated using load reduction method. In load reduction method, the

concentrated load/pressure is applied along the tunnel perimeter immediately after excavation to simulate

the initial stress state of the soil. At this stage no ground deformation is expected. This internal pressure can

be reduced with β-factor which can range between 0 and 1 to allow ground deformations or simulate the

time it takes until the secondary lining system is installed. Fig. 3 presents the load reduction factor of the

tunnel proposed by Panet and Guenot, (1982) [11]. Based on Panet and Guenot (1982) investigation, the β-

factor ranging from 20% to 80% is used by many researchers to simulate this load reduction [11].

For analyses presented in this study, the β-factors of 20% and 50% were considered. The loads along the

tunnel perimeter are reduced in three stages before the liner is installed to represent liner installation in the

field. First, the load is reduced to 80% and then to 50%. Finally, the liner is installed in third stage where

the loads are reduced to 0%.

3. RESULTS AND DISCUSSION

Results of the investigation for two analyses methods are shown in three different categories including

ground deformation in the direction of excavation, tunnel convergence and stresses transferred to the liner.

Fig. 4 and Fig. 5 illustrate tunnel crown vertical deformation versus distance along tunnel excavation for

one- and two-pass lining methods, respectively. The results were determined for five separate sequences of

excavation, ten days, one month and six months after excavation. According to the figures, crown settlement

of the tunnel increased from the first sequence to the final one gradually in both excavation methods. The

maximum settlement of the tunnel was observed at the first section of the excavation and it was

incrementally increased with tunnel advancement. In both methods, the settlement of the tunnel after ten

days was approximately equal to the final settlement of the system just immediately after final excavation

sequence (i.e. Sequence 5). The maximum settlement in one- and two-pass tunneling method was 5.8 cm

and 3.5 cm, respectively. This shows that the settlements were 40% less in two-pass lining system

comparing to one-pass lining system. It is also observed from both methods that the ground settlement

propagates about one-forth of the tunnel diameter into unexcavated portion of the tunnel along advancement

line.

The vertical ground deformations at various elevations from the tunnel crown to ground surface was

monitored chronologically for one- and two-pass lining systems are shown in Fig. 6 and Fig 7.,

respectively. The settlement profiles show that with increase in depth, vertical displacement of the ground

increased. The portion of maximum settlement of the surface to maximum vertical displacement at the

tunnel crown was approximately 14 percent for both tunneling methods within Chicago clays. This ratio is

in agreement with closed form solutions for over consolidated clays provided by Atkinson and Potts, (1977)

[12]. It is observed that at one tunnel diameter distance from the tunnel crown the vertical deformations are

reduced to 25% of maximum ground deformations.

The maximum ground surface settlement was also calculated using Cording and Hansmire (1975) methods

[13]. Based on this method for stiff clays, the inflection point of settlement trough at ground surface for a

tunnel at depth of 10.5 m and diameter of 3.8 m is 4.8 m from center line of settlement trough. The

anticipated volume loss for tunnel excavation with experienced contractor in firm ground is about 0.5%

following FHWA technical manual for design and construction of road tunnels. Consequently, the

maximum settlement at the ground surface is calculated at 0.5 cm, which closely matches with settlements

calculated in two-pass lining system (Fig. 7).

Fig. 3. Load reduction factor (after Panet and Guenot, 1982)

Fig. 4. Tunnel crown settlement in one-pass lining system

σr

0 < λ < 1

σr= (1- λ) σo

λ = 1λ = 0

σr = 0σr = σo

σo

Excavation Direction

-0.060

-0.050

-0.040

-0.030

-0.020

-0.010

0.000

012345678910

Tunnel

cro

wn v

erti

cal

def

orm

atio

ns

(m)

Distance along tunnel excavation (m)

Sequence1

Sequence 2

Sequence 3

Sequence 4

Sequence 5

10 Days

1 Month

Six Months

Fig. 5. Tunnel crown settlement in two-pass lining system

Fig. 6. Vertical ground deformation from ground surface to tunnel crown in one-pass lining method at section Y=0

-0.060

-0.050

-0.040

-0.030

-0.020

-0.010

0.000

012345678910

Tunnel

cro

wn v

erti

cal

def

orm

atio

ns

(m)

Distance along tunnel excavation (m)

Sequence 1

Sequence 2

Sequence 3

Sequence 4

Sequence 5

Ten Days

1 Month

Six Months

0

1

2

3

4

5

6

7

8

9

10

-0.060 -0.050 -0.040 -0.030 -0.020 -0.010 0.000

Dep

th f

rom

gro

und

surf

ace

till

tunnel

cro

wn

(m)

Vertical deformations (m)

Section Y=0

Sequence 1

Sequence 2

Sequence 3

Sequence 4

Sequence 5

10 Days

1 Month

6 Months

Fig. 7. Vertical ground deformation from ground surface to tunnel crown in two-pass lining method at section Y=0

Fig. 8 and 9 illustrate vertical settlement of the ground in the section, which passes through tunnel facing (i.e. Y = 5

m) for one-pass and two-pass method, respectively. The maximum settlement of the ground in one-pass and two-pass

lining system is 1.8 cm and 1.0 cm, and is observed in Sequence 5. Before Sequence 5, the deformations are typically

uniform throughout the depth and typically less than 0.5 cm. It is worth noting that the vertical deformation within 1

m above the crown is affected by the corner effect in Sequence 4; therefore, the ground deformations show a decrease

within this zone. It is also observed that the ground surface settlements at cross section Y = 0 m are almost half of

those values at cross section Y = 0 m.

Fig. 8. Vertical ground deformation from ground surface to tunnel crown in one-pass lining method at section Y=-5

0

1

2

3

4

5

6

7

8

9

10

-0.06 -0.05 -0.04 -0.03 -0.02 -0.01 0

Dep

th f

rom

gro

un

d s

urf

ace

till

tu

nn

el c

row

n (

m)

Vertical deformations (m)

Section Y=0

Sequence 1

Sequence 2

Sequence 3

Sequence 4

Sequence 5

10 Days

1 Month

6 Months

0

1

2

3

4

5

6

7

8

9

10

-0.02 -0.015 -0.01 -0.005 0D

epth

fro

m g

round

surf

ace

till

tunnel

cro

wn

(m)

Vertical deformations (m)

Section Y=-5

Sequence 1

Sequence 2

Sequence 3

Sequence 4

Sequence 5

10 Days

1 Month

6 Months

Fig. 9. Vertical ground deformation from ground surface to tunnel crown in two-pass lining method at section Y=-5

Fig. 10 and 11 demonstrate the liner stress variation along the perimeter of the tunnel obtained from one-

and two-pass lining methods, respectively. The distance along tunnel perimeter starts at tunnel crown and

extends to the lowest point of tunnel circle. It is observed that maximum stress of the liner distributed at

sides of the tunnel and it decreases to the minimum stresses at the crown and base of the tunnel. The

maximum vertical stress in two-pass lining method is 15% more than the one-pass lining method. The

stresses at the crown and base of the tunnel are almost zero in the one-pass system. This is expected as the

ground was allowed to deform due to lag in liner installation. However, in the two-pass lining system, the

stresses are larger than 10 kPa at the crown and base of the tunnel. The stress distribution along the tunnel

perimeter is uniform in two-pass lining system than the one-pass lining system. In addition, the stresses at

the liner system are gradually distributed throughout the tunneling sequences in two-pass lining system,

whereas for the other method the stress distribution is more sudden.

4. CONCLUSION

Construction of a circular tunnel was studied in the stratum of Chicago glacial clays with three dimensional

finite element numerical modeling. Two different numerical methods were selected for simulating one- and

two-pass lining system during tunnel excavation. The results of finite element analysis were provided with

the emphasis of ground stability, vertical deformation of ground and stresses transferred to the liner for both

analyses. The results indicated that maximum settlement of the ground in two-pass lining method is about

60 percent less than the maximum settlement of the ground in one-pass lining method. Furthermore, the

maximum stress of the liner was generally less and more uniform in two-pass lining system comparing to

one-pass lining method. Due to more variation in stress and ground deformations associated with one-pass

lining method, tunneling is conducted with more caution.

ACKNOWLEDGMENTS

0

1

2

3

4

5

6

7

8

9

10

-0.02 -0.015 -0.01 -0.005 0

DE

pth

fro

m g

round

surf

ace

till

tunnel

cro

wn

(m)

Vertical deformations (m)

Section Y=-5

Sequence 1

Sequence 2

Sequence 3

Sequence 4

Sequence 5

10 Days

1 Month

6 Months

The authors would like to appreciate Mr. Iman Shafii and Mr. Behrooz Moradi Bajestani, research assistants

at Southern Illinois University of Edwardsville, who helped us with this investigation.

Fig. 10. Liner stresses along the tunnel perimeter for one-pass lining system

Fig. 11. Liner stresses along tunnel perimeter for two-pass lining system

REFERENCES

1. Federal highway administration. 2013. Technical manual for design and construction of road tunnels - civil

elements. Federal highway administration.

2. Negro, A. and B.I.P. Queiroz. 2000. Prediction and performance of soft ground tunnels. Geotechnical Aspects

of Underground Construction in Soft Ground. Balkema, Tokyo, Japan. 409–418.

3. Farias, M.M., A.H.M. Junior, A.P. Assis. 2004. Displacement control in tunnels excavated by the NATM: 3-D

numerical simulations. Tunneling and Underground Space Technology. 19: 283-293.

4. Karakus, M. and R.J. Fowell. 2005. Back analysis for tunneling induced ground movements and stress

redistribution. Tunneling and Underground Space Technology. 20: 514-524.

5. Svoboda, T., D. Masin, J. Bohac. 2010. Class A predictions of a NATM tunnel in stiff clay. Computers and

Geotechnics. 37: 817-825.

6. Wang, Z., R.C.K. Wong, S. Li, L. Qiao, Finite elements analysis of long-term surface settlement above a

shallow tunnel in soft ground. Tunneling and Underground Space Technology. 30: 85-92.

7. Swoboda , G. and A. Abu-Krisha . 1999. Three-dimensional numerical modeling for TBM tunneling

in consolidated clay. Tunneling and Underground Space Technology. 14: 3, 327-333.

8. Lambrughi, A., L.M. Rodríguez, and R. Castellanza. 2012. Development and validation of a 3D

numerical model for TBM–EPB mechanised excavations. Computers and Geotechnics. 40: 97-113.

9. Finno, R.J. and M. Calvello. 2005. Supported excavations: the observational method and inverse modeling.

Journal of Geotechnical and Geoenvironmental Engineering, ASCE. 131: 826-836.

10. Finno, R.J. and C.K. Chung. 1992. Stress-strain-strength responses of compressible Chicago glacial clays.

Journal of Geotechnical Engineering, ASCE. 118: 1607-1625.

11. Panet, M. and A. Guenot. 1982. Analysis of convergence behind the face of a tunnel. Tunneling ’82, Papers

Presented at the 3rd International Symposium, Brighton, England. 197-204.

12. Cording, E.J. and W.H. Hansmire. 1975. Displacements around soft ground tunnels. 1st ed.: Storming Media. 13. Atkinson, J.H. and Potts, D.M. 1977. Subsidence above shallow tunnels in soft ground. Jnl. Geot. Eng. Div.,

ASCE, GT4, 307-325