You are on page 1of 11

Tunnelling and Underground Space Technology

Tunnelling and Underground Space Technology 20 (2005) 514524


incorporating Trenchless Technology Research

www.elsevier.com/locate/tust

Back analysis for tunnelling induced ground movements and stress redistribution
M. Karakus
a b

a,*

, R.J. Fowell

Department of Mining Engineering, Inonu University, 44280 Malatya, Turkey Department of Mining and Mineral Engineering, The University of Leeds, Leeds LS2 9JT, UK

Received 10 August 2004; received in revised form 14 February 2005; accepted 16 February 2005 Available online 3 June 2005

Abstract Analysing tunnelling process in 2D plane strain conditions is widely used method to calculate tunnelling induced settlement proles as well as soil structure interactions. Possibility of damage to the surface and/or underground structures can be estimated using powerful nite dierence method (FDM) and nite element method (FEM) of analysis. However, setting up a realistic model that would be able to achieve this goal is rather dicult. In this paper, 2D FDM analysis has been conducted to assess tunnelling induced settlement, stress redistribution phenomena along with movements around shallow soft ground tunnels excavated in accordance with the New Austrian Tunnelling Method. Measurements recorded during construction of the Heathrow Express Trial Tunnel in London Clay were compared with the predicted values to validate numerical estimations. As a soil model, the MohrCoulomb plasticity model has been used in the FDM analysis. Results obtained from 2D FEM are also included in this paper for comparison purposes to evaluate performance of both numerical analysis procedures. Predictions from both FDM and FEM analyses proved to be procedures used within this work can be a tool in practical engineering applications to simulate tunnelling operations. 2005 Elsevier Ltd. All rights reserved.
Keywords: The Heathrow Express Trial Tunnel; FEM; FDM; London Clay; Surface and sub-surface settlement; Cable elements; Stress redistribution; NATM

1. Introduction Overpopulation leading to an increase in the number of commuters in urban areas has caused acceleration in constructing underground Metro tunnels to overcome transportation problems. Therefore, all the parameters having inuences on the magnitude and the proles of the surface settlement have to be investigated thoroughly as the major concern in constructing such tunnels in urban areas is to reduce and/or to control settlement. Using numerical analysis can often predict the consequences without using any full scale trial tunnels.
*

Corresponding author. Tel.: +90 422 341 00 30; fax: +90 422 341 00 E-mail address: mkarakus@inonu.edu.tr (M. Karakus).

46.

There are empirical models developed to estimate surface settlements in both transverse and longitudinal directions to the tunnel axis (Peck, 1969; New and OReilly, 1978; OReilly, 1988; Mair et al., 1993; New and Bowers, 1994). These empirical models produce very good surface settlement predictions. However, considering that the empirical models are mainly based on past experience, these models are conservative for a ground that has not had any tunnelling process conducted in. In the preliminary tunnel design stage, not only employing correct soil stiness properties but also selfweight of structures on the surface to numerical models can predict the probability of damage to surface structures. Therefore, there have been large numbers of numerical analyses in 2D and 3D conducted to assess tunnelling induced damage (Swoboda, 1979; Gunn,

0886-7798/$ - see front matter 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.tust.2005.02.007

M. Karakus, R.J. Fowell / Tunnelling and Underground Space Technology 20 (2005) 514524

515

1992; Shohrour and Ghorbanbeigi, 1994; Swoboda et al., 1994; Dasari, 1996; Swoboda et al., 1999; Mroueh and Shahrour, 2003; Karakus and Fowell, 2003). In the present work, 2D plane strain Finite Dierence analyses have been conducted to investigate ground movement proles and stress redistribution around a NATM tunnel. Fast Lagrangian analysis of continua (FLAC) (Itasca, 1993) program has been utilized to simulate tunnel construction. Tunnel Excavation process has been modelled using the hypothetical modulus of elasticity (HME) soft lining approach (Powell et al., 1997). Mroueh and Shahrour (2003) has studied interactions of buildings with tunnels using 3D nite element analysis. They concluded that self weight of buildings on the surface has a major role on determination of initial stresses in the ground and neglecting this results in underestimation of tunnelling induced forces leading to less settlement predictions. Therefore, surcharge of 80 kPa (Bowers, 1997) due to existing car park over the Heathrow Trial Tunnel excavation has been included within the analysis. Surface and subsurface transverse settlement troughs, along with horizontal stress distributions measured during the Heathrow Trial Tunnel construction in 1992 have been compared with the predicted results (Bowers, 1997).

eective stress values eective Young modulus (E 0 ) and eective poissons ratio (v 0 ) (Burland and Kalra, 1986). Made ground and Terrace gravel (Thames gravel) were modelled using the drained material properties. Properties adopted for the numerical analysis are given in Table 1. Grose and Eddie (1996) suggested that the cohesion of the London Clay is 10 kPa and earth pressure at rest, K0 = 1 when analysing the Heathrow Transfer Baggage System tunnel. On the other hand, it was found by Atzl and Mayr (1994) and Ryley and Carder (1995) that the cohesion of the London Clay varies between 0 and 30 kPa. Earth pressure at rest (K0) value for made ground and Thames gravel was predicted using well known Jakys formula (Konc = 1 sin / 0 ) and adopted in the FDM analysis (see Fig. 1). According to the data derived from dierent sites in London, Burland and Kalra (1986) proposed a relation for the drained Young Modulus of London Clay in both vertical and horizontal directions. These relationships show that stiness of the London Clay varies with depth (z). E0m 7.5 3.9z MN=m2 ; E0h 1.6E0m MN=m2 . 1 2

2. Geology at the site and construction processes Materials encountered at the site consist of Thames gravel, made ground and London Clay. The London Clay is clearly dominant at the site and is found below a depth of 4.2 m. This is overlain by coarse gravel, with 0.3 m of cement-stabilised material and above this a bitumen covered car park (Ryley and Carder, 1995). Dean and Basset (1995) reported that London Clay is generally homogenous with very few major discontinuities and it has a good stand up time of at least 18 h. Trial tunnel construction was carried out approximately 16.8 m crown depth. The analyses were carried out in terms of eective stress parameters, in other words, drained analyses were conducted. Therefore, London Clay properties were expressed in terms of equivalent
0
Depth below clay surface (z),m

Values of the London Clay Modulus, MPa 0 50 100 150 200 250

10 20 30 40 50 60

E'v (Burland, 1986) Bulk Modulus (K) Shear Modulus (G)

Fig. 1. Variation of London Clay Modulus with depth adopted in FLAC analysis (Burland and Kalra, 1986).

Table 1 Drained material properties at the site (Powell et al., 1997; Burland and Kalra, 1986) Parameters, units and symbols Bulk unit weight, kN/m , c Cohesion, MPa, c 0 Earth pressure at rest, K0 Internal friction angle, / 0 () Dilation angle, u 0 () Poissons ratio, m 0 Eective bulk modulus, MPa, K 0 Eective shear modulus, MPa, G 0 Eective Youngs modulus (in vertical), MPa, E0m
3

London Clay 20 10 1.15 25 12.5 0.125 See Fig. 1 See Fig. 1 See Fig. 1

Thames Gravel 19 0 0.43 35 17.5 0.3 23.8E3 19.23E3 50E3

516

M. Karakus, R.J. Fowell / Tunnelling and Underground Space Technology 20 (2005) 514524

Construction of the trial tunnel started in February 1992 with the Type-1 (TS1) method of advance, a double side drift sequence, followed by Type-2 (TS2), single side drift sequence, and Type-3 (TS3), crown, bench, and invert face excavation (Fig. 2). This work was completed by 16 June 1992. Measurement and monitoring of the construction process was conducted by the Transport Research Laboratory (TRL). The project provided information on ground movements in the London Clay to tunnel designers. Each trial section progressed for at least 30 m in order to obtain adequate and meaningful data over each section. Trial tunnel construction was carried out at 16.8 m crown depth below the surface with having 7.9 m height and 9.2 m width, producing a 100-metre long running tunnel. From the eld observations, Type-2 (TS2) produced the minimum transverse surface settlement prole among the Trial tunnels and TS2 was subjected to the FDM analysis in this research. The shotcrete has been represented by elastic beam elements in the analysis. The elasticity modulus of the beam elements was divided by (1 m2) in order to take account the plane strain conditions since beam element formulation is a plane stress formulation implemented in the FLAC. Beam elements attached to the ground and cable elements were used to simulate rock bolts. Properties of these structural elements are given in Table 2. Cable elements are 1D axial structural elements anchored at a specic point in the grid or grouted so that

the cable element develops forces along its length as the grid deforms. Cable elements are used to model rock bolts, cable bolts. Cable element formulation in FLAC considers deformations along entire length of bolts rather than a specic point and thus it is useful modelling such reinforcement systems as rock bolts where grout material may fail in shear in some length of the reinforcement (Itasca, 1993). Fig. 3 illustrates the cable element behaviour implemented in FLAC. As the input parameters for cable elements, grout stiness (Kbond), grout shear strength (Sbond), and area (A), Youngs Modulus (E), diameter (D), length (L) and ultimate strength of cable are required. A 1D constitutive model is used to model axial behavior of the reinforcing element in FLAC. The axial stiness is described in terms of the reinforcement cross-sectional area, A, and Youngs modulus, E(w), the incremental axial force, DFt, is calculated from the incremental axial displacement by DF t EA t Du ; L
b a b a

where Dut u1 u1 t1 u2 u2 t2 . The superscripts [a], [b] refer to the nodes in the grid. The cosines t1, t2 are the tangential (axial) direction of the cable. As a consequence of relative shear displacement, ut, between the tendon surface and the borehole surface, the shear force, Ft, mobilized per length of cable is related to the grout stiness, Kbond i.e. (Itasca, 1993), F t K bond ut . 4

Fig. 2. The Heathrow Express Trial Tunnel, Type-2 (Bowers, 1997). Table 2 Structural element properties used in the FDM analysis (Itasca, 1993; Bowers, 1997) Parameters, symbols and units Area, A, m2 Poissons ratio, m Youngs modulus, E, MPa Moment of inertia, I, m4 Cable diameter, D, m Cable length, L, m Ultimate strength of cable (force), MN Bond stiness of grout, Kbond, MN/m/m Bond strength of grout, Sbond, MN/m n/a: not applicable. Beam elements Inner lining 0.15 0.15 5000 0.000281 n/a n/a n/a n/a n/a Outer lining 0.25 0.15 5000 0.0013 0.0005 40000 0.025 3 0.225 6000 0.320 Cable elements

M. Karakus, R.J. Fowell / Tunnelling and Underground Space Technology 20 (2005) 514524

517

Fig. 3. Cable element behaviour in FLAC (Itasca, 1993).

was measured using spade cells and push-in soil stress cells that were inserted near to the tunnel crown and at axis level. To measure the horizontal total stresses, these cells were incorporated with the pneumatic piezometers to record pore water pressures. Pneumatic piezometers were installed in sand cells within the spade cell boreholes with bentonite plugs above and below. Three extensometers were placed over the axis perpendicular to the Tunnel Type-2 and one on the tunnel centreline. The magnetic rings located at various depths, which were ranged from 3 to 27 m depth. Biaxial inclinometers with automatic loggers were used to record the horizontal movements using techniques developed by TRL (Bowers, 1997). A detailed location drawing for the inclinometers are given in Fig. 4. As in-tunnel instrumentation, convergence-measuring pins were used to measure horizontal and vertical movements of the shotcrete with time. These were installed on the tunnel periphery when the shotcrete was installed. 2.2. General numerical analysis procedure MohrCoulomb plasticity model has been adopted throughout the FLAC analysis to represent stressstrain behaviour of not only London Clay but also Thames gravel and made ground. MohrCoulomb criterion has a linear failure surface corresponding to shear failure as described in the following equations: p fs r1 r3 N / 2c N / ; 9 N / 1 sin /=1 sin /; 10

If laboratory pull-out tests are available, Kbond can be measured directly, else the stiness can be calculated from a numerical estimate for the elastic shear stress, sG, obtained from an equation describing the shear stress at the grout/rock interface (Itasca, 1993): G Du ; 5 sG D=2 t ln1 2t=D where Du is relative displacement between the element and the surrounding material; Gw is grout shear modulus; Dw is reinforcing diameter; and tw is the annulus thickness. Consequently, the grout shear stiness, Kbondw ( N/ m/m w, is simply given by 2pG . 6 K bond ln1 2t=D Other important parameter which has to be dened is Sbond. This may also be deduced from the following expressions when ignoring frictional connement eects (Itasca, 1993): S bond pD 2tspeak ; speak sI QB ; 7 8

where r1 is major principal stress; r3 is minor principal stress; / is friction angle and c is the cohesion. Shear yield is detected if fs<0 (Itasca, 1993). For modelling of the tunnelling process in 2D, an approach has to be adopted to take into account the deformation occurring prior to shotcrete installation and the

where speak is the maximum shear force per cable length in the grout, sI is about 50% of the uniaxial compressive strength of the weaker of the rock and grout, and QB is the quality of the bond in between grout and rock (QB = 1 is for perfect bonding) (Itasca, 1993). 2.1. The Heathrow Express Trial Tunnel Type-2 and instrumentation methodology Sub-surface measuring equipment consists of extensometers, inclinometers and push-in pressure cells and piezometers. Magnetic probe extensometers were used to measure vertical displacement. Horizontal soil stress

Fig. 4. Surface, subsurface and in-tunnel instrumentation around the Heathrow Trial Tunnel Type-2 (Bowers, 1997).

518

M. Karakus, R.J. Fowell / Tunnelling and Underground Space Technology 20 (2005) 514524
100m

3 2

50 m

1-Outer lining300 mm, 2-Inner lining250 mm, 3-Cablebolts

Fig. 5. Model grid used in the FLAC.

3D tunnelling problem. There are dierent approaches to consider aforementioned deformations such as the convergence connement method (Panet and Guenot, 1982), the progressive softening method (Swoboda, 1979), volume loss control method (Potts and Zdravkovic, 2001). The hypothetical modulus of elasticity (HME) soft lining approach, a relatively new technique developed for the simulation of Heathrow Express tunnel at Terminal 4 by Powell et al. (1997), has been employed in the analysis. Essential to this approach is to introduce a lower elasticity modulus for the shotcrete to represent the excavation process and subsequently increase it to the assumed short-term modulus of elasticity for the shotcrete. The approach is believed to model the shotcrete behaviour until it gains strength. More detailed explanation of this approach is given by Karakus (2000) and Karakus and Fowell (2003). The model grid used in the analysis is shown in Fig. 5. The model was xed in the horizontal direction at each side, which means that vertical movement was allowed,
Offset from tunnel centreline, m
50 30 10 0 -10 -30 -50

and the bottom part of the boundary was pinned, so neither vertical nor horizontal movements were allowed. As can be seen in Fig. 5, the top surface of the model was free in both directions. The construction process of the Trial Tunnel Type-2 consists of a left hand heading, bench and invert excavation. This was followed by the excavation of the right hand heading, bench and nally closing the ring of support. In order to imitate the same construction process in FLAC analysis, sequential excavation model (SEM) was employed (Karakus and Fowell, 2003). The SEM excavation process was modelled with the following main stages: 1. Establishing the equilibrium condition for the model body by setting up a gravitational stress eld and introducing 80 kPa surcharges owing to the car park.

Offset from tunnel centreline, m


50 40 30 20 10 0 0 5 -10 -20 -30 -40 -50

Surface settlement, mm

Surface settlement, mm

10 15 20 Sidewall with cable Sidewall without cable Enlargement with cable 30 35 25

10

15

20

FLAC_SEM sidewall Measurements sidewall FLAC_SEM enlargement Measurements enlargement

25

30

Fig. 6. Predicted surface settlement proles by FDM.

Fig. 7. Predicted surface settlement with and without cable elements by FDM analysis.

M. Karakus, R.J. Fowell / Tunnelling and Underground Space Technology 20 (2005) 514524

519

Horizontal Displacement, mm
0 10 20
0

20

40

-5 0

10

15

-40

-20
IB4 (-7.69m)

0 0

0 5 10 15 20 25

IB1 (10.04m) 5
10 15 20 25 30

IB2 (7.62m)

5 10 15 20 25 30 35

IB3 (4.35m)

5 10

C
16.8m

15 20

Tunnel

25
FLAC predictions Field measurements

30
35

30 35

35

Fig. 8. Predicted horizontal movements compared to eld measurements.

Subsurface settlement, mm
0 0 5 Depth, m 10 15 20 25 30 3 6 9 12 15 18
0 5 10 15 20 25 30 0 3

Subsurface settlement, mm
6 9 12 15 18 21

FLAC MB1 Measurement MB1

FLAC MB2 Measurement MB2

Subsurface settlement, mm
-5 0 5 FLAC MB3 Measurement MB3 0 5 10 15 20 25 30
0 5 10 15 20 25 30 -10 -5 0

Subsurface settlement, mm
5 10 15 20 25 30

FLAC MB4 Measurement MB4

Depth, m

10 15 20 25 30

Subsurface settlement, mm
0 0 2 4 6 8 10 12 14 16 18 10 20 30 40 50 60

Depth, m

FLAC MT1 Measurement MT1

Fig. 9. Predicted subsurface settlements proles.

Depth, m

520

M. Karakus, R.J. Fowell / Tunnelling and Underground Space Technology 20 (2005) 514524

Surface settlement, mm

2. Having reached equilibrium conditions, elements in the left hand heading were removed, followed by activation of the beam elements in this section with a lower elasticity modulus of 0.40 GPa assigned which was found from the back analysis. 3. Removing elements in the left hand bench and invert and activating the beam elements for this section with an elasticity modulus of 0.40 GPa. At the same time the HME value of the left heading beam elements was increased to 5 GPa, assumed to be the short-term modulus of the shotcrete. 4. The same procedures in stage 2, and stage 3 have been applied to the right hand heading, bench and invert excavations. 5. Removing the inner wall was the nal step of the FDM analysis. 0.40 GPa of HME value found from back analysis. According to Karakus and Fowell (2003), HME value can be found from the following expression which is similar to the progressive softening method (Swoboda, 1979): HME dEShort
term

crete. Total amount of reduction in stiness of shotcrete is (1 d) which is 92% in this case. This value indicates that beam elements used to represent shotcrete behaves in a very sti manner as supported by Augarde and Burd (2001) who reported that structural elements such as shell elements to model liner may behave in an oversti manner when embedded in a mesh of continuum element.

Offset from tunnel centreline, m


50 40 30 20 10 0 0 -10 -20 -30 -40 -50

10

15

20 ABAQUS sidewall FLAC sidewall Measurements sidewall ABAQUS enlargement FLAC enlargement Measurments enlargement

11

Cable elements were included in FLAC analysis

25

30

where d is the reduction factor as percentage and the EShort term is the Short term Youngs modulus of shot-

Fig. 11. Predicted transverse surface settlement proles by FLAC and ABAQUS.

700 600 500 Stress, kPa 400 300 200 100 0 04/11/92 04/18/92 04/25/92 05/02/92 05/09/92 05/16/92 Dates 04/11/92 04/18/92 04/25/92 05/02/92 05/09/92 05/16/92 500 400 300 Stress, kPa Stress, kPa 200 100 0 -100 -200 -300 FLAC SB3 Measurement SB3 Stress, kPa FLAC SB1 Measurement SB1

800 700 600 500 400 300 200 100 0 04/11/92 04/18/92 04/25/92 05/02/92 05/09/92 05/16/92 Dates 04/11/92 04/18/92 04/25/92 05/02/92 05/09/92 05/16/92 600 500 400 300 200 100 0 FLAC ST 1 Measurement ST 1 FLAC SB2 Measurement SB2

Fig. 10. Predictions for the horizontal stress redistribution around the tunnel.

M. Karakus, R.J. Fowell / Tunnelling and Underground Space Technology 20 (2005) 514524

521

3. Evaluation of the results 3.1. Transverse surface settlement analysis Results of the analyses have been shown in three main parts as surface settlement proles, sub-surface settlement proles and stress redistribution around tunnel. Surface settlement analysis has been carried out in accordance with the eld measurements corresponding to the completion of the left sidewall excavation and the enlargement of the tunnel as illustrated in Fig. 6. Karakus and Fowell (2003) reported that 0.40 GPa of HME value produced the best results for sequential excavation model. Thus, for this analysis, 0.40 GPa value was taken as HME value. As can be seen from Fig. 6, surface settlement proles for both sidewall excavation and enlargement of the tunnel are in very close agreement with the eld measurements. Maximum surface settlement predicted by the SEM model for sidewall excavation and enlargement are 15.3 and 27.6 mm, respectively. Including cable elements during numerical analysis prevented excessive settlement at the either side of the tunnel headings as shown in Fig. 7. Cable elements included in the analysis and assumed to be

fully grouted. The material properties adopted for the cable elements and beam elements are summarised in Table 2. Fig. 7 illustrates the eects of cable elements on the surface settlement. The settlement proles were not affected 1.5-diameter away from either side of the tunnel centreline in both including and excluding cable elements in the numerical analysis. However, the settlement proles for the sidewall and the enlargement of the tunnel are greatly inuenced within 1.5D of either side of the tunnel when no cable elements were present. Therefore, the use of cable elements reduced the ground movements within 1.5D of either side of the tunnel. Without cable elements, approximately 2 and 4 mm more surface settlements have been predicted for sidewall and enlargement, respectively. Hence, the use of cable elements for NATM analysis is considered important. 3.2. Subsurface settlement analysis The horizontal movements and subsurface settlement predictions from the FLAC are compared with the corresponding inclinometers measurements and magnetic ring measurements, respectively. Thus, these comparisons provide a verication of the predictions obtained from the nite dierence analysis. Fig. 8 illustrates the calculated horizontal movements towards the tunnel. Although calculated movements for the IB3 was two times greater than the measured values, predicted horizontal movements are in good agreement with the inclinometer measurements IB1, IB2, and IB4. Besides, predictions corresponding to IB1 and IB2 became greater initiating from a depth of 15 m to the ground surface. On the other hand, predicted subsurface settlements for all magnetic measurements are in very good agreement with the measurements (Fig. 9).

Table 3 Comparison of the maximum surface settlements Numerical models used Maximum surface settlement at tunnel centreline Sidewall (mm) Field measurement FLAC with cable elements FLAC without cable elements ABAQUS without cable elements 14.6 15.3 17.3 11.4 Enlargement (mm) 26.8 27.6 31.4 27.4

Horizontal Displacement, mm

0 0 5 10 15 20 25 30 35

10

20

0 0

0
IB2 (7.62m)

40 -10

0 0 5 10 15 20 25 30 35

10

20
IB3 (4.35m)

-40

-20

0 0

IB1 5 (10.04m)

IB4 (-7.69m)

5 10

10 15 20 25 30 35

16.8m

15 20

Tunnel
FLAC predictions ABAQUS predictions Field measurements

25 30 35

Fig. 12. Comparison of FLAC and ABAQUS predictions for horizontal movements corresponding to inclinometer measurements.

Depth, m

522

M. Karakus, R.J. Fowell / Tunnelling and Underground Space Technology 20 (2005) 514524

This is believed to be due to adopting variation of London clay Elasticity modulus with depth, which was suggested by Burland and Kalra (1986), in the FLAC analysis.

General trend of stress changes for both FLAC predictions and the eld measurements can be explained as follows: 1. When the rst heading is excavated, the horizontal stresses are decreasing sharply from the geostatic stress condition to a lower stress level. 2. Then, the application of the shotcrete stabilises the stress distribution around the tunnel until a new section was excavated. This phenomenon continues until the entire tunnel construction is completed and the support ring is closed. Calculated horizontal stresses are in close agreement with the spade cells measurements, SB1, SB2, and ST1, while predictions for the spade cell SB3 is greater

3.3. Stress analysis around the tunnel The locations of the spade cells are illustrated in Fig. 4. The real time of construction was not considered in the analysis. However, the stresses at the end of each main excavation sequences were used for comparison with the corresponding eld measurements. Thus, only related measurements were used in this analysis. Fig. 10 shows the horizontal stress redistribution around the tunnel.

0 0 5 10

Subsurface settlement, mm 6 9 12 FLAC MB1 ABAQUS MB1 Measurement MB1

15

18 0 5 10 15 20 25 30 35

Subsurface settlement, mm 6 9 12 15

18

21

FLAC MB2 ABAQUS MB2 Measurement MB2

Depth, m

15 20 25 30 35 Subsurface settlement,mm 15 25

-5 0 5 10

35

-10 0

Subsurface settlement, mm 10 20

30

40

FL AC MB3 ABAQUS MB3 Measurement MB3

5 10 15 20 25 30

FLAC MB4 ABAQUS MB4 Measurement MB4

Depth, m

15 20 25 30

0 0 2 4 6

10

Subsurface settlement, mm 20 30 40

50

60

Depth, m

8 10 12 14 16 18 FLAC MT 1 ABAQUS MT 1 Measurement MT 1

Fig. 13. Comparison of predicted subsurface settlement proles with magnetic ring measurements.

M. Karakus, R.J. Fowell / Tunnelling and Underground Space Technology 20 (2005) 514524
700 600 500
FLAC SB1 ABAQUS SB1 Measuremenst SB1

523

800 700 600 500


FLAC SB2 ABAQUS SB2 Measurement SB2

Stress, kPa

400

400
300 200 100

300 200 100

0 0 04/11/92 04/18/92 04/25/92 05/02/92 05/09/92 05/16/92 04/11/92

04/18/92

04/25/92

05/02/92

05/09/92

05/16/92

Dates
04/11/92 04/18/92 600 500 400
FLAC SB3 ABAQUS SB3 Measurement SB3

Dates
04/18/92 04/25/92 05/02/92 05/09/92 05/16/92 600 500 400 300 200 100
FLAC ST 1 ABAQUS ST 1 Measurement ST 1

04/25/92 05/02/92 05/09/92 05/16/92 04/11/92

Stress, kPa

300 200 100 0 -100 -200

Fig. 14. Comparison of FLAC and ABAQUS stress predictions corresponding to spade cell measurements.

than the measured stress. Similar predictions were found for SB3 with the nite element analysis (Karakus and Fowell, 2003). This implies that the accuracy of the spade cell measurement SB3 might not be as accurate as the other spade cell measurements. 3.4. Comparison of FDM and FEM analysis results In this section, overall comparison of the nite element and nite dierence analysis, is made so that the dierences between these two powerful modelling tools can be examined. In addition, this comparison will provide an insight into the eects of using dierent plasticity models for London Clay. For the Finite element analysis, ABAQUS program has been utilized detailed analysis procedure is given elsewhere (Karakus and Fowell, 2003). During FEM analysis, London Clay stress strain behaviour was represented by Modied Cam-clay plasticity with porous elasticity. The DruckerPrager plasticity model was used for Thames gravel and made ground. As can be seen from Fig. 11 and Table 3, surface settlement predictions by FLAC and ABAQUS for the tunnel enlargement are in very close agreement with each other and with eld measurements as well. Conversely, FDM analysis produced better settlement predictions for sidewall excavation. This is believed to be due to variation of London Clay elasticity modulus with depth, which was considered in the FDM analysis.

Horizontal displacements towards tunnel and subsurface settlements proles are illustrated in Figs. 12 and 13, respectively. Finite element and nite dierence predictions for horizontal movement are in good agreement with each other. Nevertheless, FDM analysis failed to predict the horizontal movement 4.35 m away from the tunnel centreline. Also, FDM analysis overestimated these movements for IB1 and IB2 from above 15 m depth towards the ground surface. This can be attributed to the use of dierent soil models considered for both analyses. Subsurface settlement calculated from both models is well matched to the eld measurements. Apparently, the contradictions between surface settlement and horizontal movements from both analyses show that anisotropy in the London clay should be considered in the numerical analysis. However, plasticity models especially the Modied Cam-clay model did not take into account anisotropy in ABAQUS program as it was originally developed for isotropic conditions. Stress redistribution around the tunnel predicted by FDM and FEM are in agreement with the eld measurements. However, predictions obtained from the nite element analysis are much closer to the eld measurement than the nite dierence analysis predictions. Both analyses have failed to predict the horizontal stresses for the SB3 spade cell measurement as can be seen from Fig. 14. As explained earlier, this could be due to failure in spade cell measurements. Bowers (1997) reported that

524

M. Karakus, R.J. Fowell / Tunnelling and Underground Space Technology 20 (2005) 514524 Bowers, K.H., 1997. An appraisal of the New Austrian Tunnelling Method in soil and weak rock. Ph.D. Thesis, The University of Leeds, 240 pp. Burland, J.B., Kalra, J.C., 1986. Queen Elizabeth II Conference Centre: geotechnical aspects. Proc. Inst. Civil Eng. 80 (1), 1479 1503. Dasari, G.R., 1996. Numerical modelling of a NATM tunnel construction in London Clay. In: International Symposium on Geotechnical Aspects of Underground Construction in Soft Ground. Balkema, Rotterdam, pp. 491496. Dean, A.P., Basset, R.H., 1995. The Heathrow Express Trial Tunnel. Proc. Inst. Civil Eng., Geotech. Eng. 113, 144156. Grose, W.J., Eddie, C.M., 1996. Geotechnical aspects of the construction of the Heathrow Transfer Baggage System tunnel. In: International Symposium on Geotechnical Aspects of Underground Construction in Soft Ground. Balkema, Rotterdam, pp. 269276. Gunn, M.J., 1992. The prediction of surface settlement proles due to tunnelling. In: Proceedings of the Wroth Memorial Symposium held at St. Catherines College, Oxford, pp. 304316. Itasca Consulting Group Inc., 1993. FLAC Users Manual I, Minneapolis, MN. Karakus, M., 2000. Numerical modelling for NATM in soft ground. Ph.D. Thesis, The University of Leeds, 240 pp. Karakus, M., Fowell, R.J., 2003. Eects of dierent tunnel face advance excavation on the settlement by FEM. Tunnelling Underground Space Technol. 18 (5), 513523. Mair, R.J., Taylor, R.N., Bracegirdle, A., 1993. Subsurface settlement proles above tunnels in clays. Geotechnique 43 (2), 315320. Mroueh, H., Shahrour, I., 2003. A full 3-D nite element analysis of tunnelling adjacent structures interaction. Comput. Geotech. 30, 245253. New, B.M., Bowers, K.H., 1994. Ground movement model validation at the Heathrow Express trial tunnel. In: Tunnelling94, IMM, London, pp. 301329. New, B.M., OReilly, M.P., 1978. Tunnelling induced ground movements; predicting their magnitude and eects. In: Proceedings of the Conference on Large Ground Movements and Structures. Pentech Press, Cardi, pp. 671693. OReilly, M.P., 1988. Evaluating and predicting ground settlements caused by tunnelling in London Clay. In: Tunnelling88, IMM, London, 1988, pp. 231241. Panet, M., Guenot, A., 1982. Analysis of convergence behind the face of a tunnel. In: Tunnelling82, IMM, London, pp. 197204. Peck, R.B., 1969. Deep excavations and tunnelling in soft ground. In: Proceedings of the 7th International Conference on Soil Mechanics and Foundation Engineering, Mexico, pp. 225290. Potts, D.M., Zdravkovic, L., 2001. Finite Element Analysis in Geotechnical Engineering Application. Thomas-Telford, London, 427 pp. Powell, D.B., Sigl, O., Beveridge, J.P., 1997. HeathrowExpress-design and performance of platform tunnels at Terminal 4. In: Tunnelling 97, IMM, London, pp. 565593. Ryley, M.D., Carder, D.R., 1995. The performance of push-in spade cells installed in sti clay. Geotechnique 45 (3), 533539. Shohrour, I., Ghorbanbeigi, S., 1994. Calculation of tunnels in soft ground. In: Numerical Methods in Geotechnical Engineering. Balkema, Rotterdam, pp. 229234. Swoboda, G., 1979. Finite element analysis of the New Austrian Tunneling Method (NATM). In: Proceedings of the 3rd International Conference on Numerical Methods in Geomechanics, Aachen, vol. 2, pp. 581586. Swoboda, G., Marence, M., Mader, I., 1994. Finite element modelling of tunnel excavation. Int. J. Eng. Modell. 6, 5163. Swoboda, G., Ichikawa, Y., Dong, Q., Zaki, M., 1999. Back analysis of large geotechnical models. J. Numer. Anal. Meth. Geomech. 23, 14551472.

some of the stress cells failed before tunnelling was started. According to both numerical analyses results, there would be a possibility of damage in this particular stress cell, SB3.

4. Conclusions HME approach which is used in both FDM and FEM analysis to account for volume loss during tunnelling operations proved to be a practical tool for numerical modeller. The 0.40 GPa of HME value found from back analysis of FEM produced very good predictions in FDM analysis as well. As dierent soil plasticity models considered viz. Modied Cam-clay and MohrCoulomb plasticity for FEM and FDM, respectively, it can be proposed that HME approach can be used in which dierent plasticity models apart from above ones considered in a numerical analysis. The results of present analyses showed that both methods of analyses using the sequential excavation model could be used for the preliminary design of NATM tunnelling for the conditions used in this research. It was found that the variation of elasticity modulus of London clay with depth is very important for accurate numerical modelling. Thus, FDM analysis predicted much closer surface settlement for the sidewall excavation than the nite element predictions. Contradictions between subsurface settlement and horizontal displacements suggest that considering anisotropic behaviour of London clay could improve predictions in both FEM and FDM analyses. Thus, this problem should be subjected to further analysis to examine their eects on NATM tunnelling. FDM analysis has also shown the importance of cable elements adopted in the numerical analysis. Without considering cable elements in the numerical analysis led to greater settlements.

Acknowledgments Grateful acknowledgement is given to Inonu University, Turkey for providing nancial support during the research undertaken in the Department of Mining and Mineral Engineering at the University of Leeds, UK.

References
Augarde, C.E., Burd, H.J., 2001. Three-dimensional nite element analysis of lined tunnels. Int. J. Numer. Anal. Meth. Geomech. 25, 243262. Atzl, G.V., Mayr, J.K., 1994. FEM-analysis of Heathrow NATM trial tunnel. In: Numerical Methods in Geotechnical Engineering. Balkema, Rotterdam, pp. 195201.

You might also like