[doi: 10.1680/geot.2008.58.9.

693]

Introduction for the 46th Rankine Lecture

The 46th Lecture of the British Geotechnical Society was given by Professor Robert J. Mair at Imperial College London on 22 March 2006. The following introduction was given by Professor R. N. Taylor, City University Mr Chairman, distinguished guests, ladies and gentlemen, it is both an honour and an immense pleasure to introduce this evening Professor Robert Mair, the British Geotechnical Associations 46th Rankine Lecturer. Robert James Mair was born in 1950 and brought up in Cambridge, a place that has had a deep inuence on his life and career. He read engineering at Clare College, and was inspired by the lectures on soil mechanics given by Peter Wroth. On graduating in 1971 he was determined to work for a company with geotechnical engineering prominent in its portfolio. That company was Scott Wilson Kirkpatrick and Partners, and it was a happy coincidence that on his rst day in their ofces he was given a desk next to David Hight, who was to become one of Roberts closest colleagues and inuential friends. (And I would like to say now that David is very disappointed to have been called overseas and is not able to be here today.) It was not too long before Robert moved to the Hong Kong ofce of SWK, and in those heady days experienced a truly dynamic engineering environment, which proved to be a great beginning for an engineer fully tuned to learning quickly. Roberts venture to Hong Kong came to an end with an offer from Andrew Schoeld to undertake research at Cambridge University. Scott Wilson Kirkpatrick agreed to second Robert to the Cambridge Soils Group, where major research was being supported by Myles OReilly of the then Transport and Road Research Laboratory, who was keen to get some science into the understanding of tunnel stability and the development of ground movements. It started Roberts professional association with tunnelled excavations for which he is now world renowned. Robert became the natural leader of the informal tunnelling and buried pipes research group at Cambridge, and created a strong sense of team responsibility, leading to sound, well-organised research. I count myself fortunate to have become part of that group, and to have got to know Robert at that time. They were great times, with Robert clearly the team leader, working hard to get the most out of people but also creating a fun atmosphere, including arranging cricket competitions with the Imperial College soil mechanics group. Roberts own research made use of the then relatively new technique of geotechnical centrifuge modelling. This resulted in seminal research on tunnel stability, perhaps remaining the best illustration of centrifuge modelling applied directly to geotechnical design, and which is still regularly cited today. Robert returned to Scott Wilson Kirkpatrick in 1980, but it was not too long before he was seeking new challenges. In 1983 an opportunity arose for Robert to join forces with David Hight and Peter Vaughan, both then of Imperial College, to create a specialist geotechnical ofce known to us all now as the Geotechnical Consulting Group. There was no certainty then that this adventure could ever be successful, but we have witnessed the company grow from its origins in the small two-room ofce in Kendrick Mews to what is now one of the worlds leading specialist geotechnical companies. It has always been a pleasure to visit the GCG ofces. I nd there a convivial atmosphere that so strongly reects Roberts character, and have enjoyed many a picnic lunch in the kitchen, engaging in all 693

Professor R. J. Mair

manner of technical, sporting, political or personal discussions. Roberts professional expertise has seen him involved with a wide range of projects, and he has accomplished a great deal. With Chris Padeld he produced CIRIA Report 104 on the Stability of retaining walls, which became one of CIRIAs most used publications and was hugely inuential. He was closely involved with the Jubilee Line Extension Project, Crossrail, the Channel Tunnel Rail Link, London Undergrounds Angel Station reconstruction project and the Waterloo escalator tunnel, the last leading to the development and use of a new technique for the control of tunnelinduced ground movements termed compensation grouting, a phrase coined by David Hight. In fact there is barely a tunnel in London that does not have Roberts name on it. Roberts breadth of expertise, sense of organisation and diplomatic skills have made him a natural choice for many expert review panels, advisory panels and other committees including, importantly for me, the Board of the International Society for Soil Mechanics and Geotechnical Engineering. Throughout, Robert has retained an interest in teaching and education. He was a key contributor to the short courses on tunnelling in soft ground given at City University, and in 1997 became Royal Academy of Engineering Visiting Professor at Cambridge University. The lure of Cambridge was again taking hold, and in 1998 he moved back there, this time as Professor of Geotechnical Engineering. He quickly took charge of the Soil Mechanics Group and instilled a strong team spirit, ensuring a well-organised, highly motivated and committed research group. His own research projects focused on practical geotechnical engineering problems, and allowed the Centre for Construction Processes to be created in the Schoeld Centre at the west Cambridge

694

MAIR of people, immensely supportive, and always prepared. To these talents he adds those of a fantastic mimic and regularly does takes on many characters, a particular speciality being his wonderful impersonation of the late Bill Ward. He is a brilliant communicator, and has a talent for simplifying problems while retaining the key issues. He has an undying enthusiasm for the subject of soil mechanics and its practice in engineering, and he is genuinely interested in understanding and solving problems that the ground throws up. With this he has established a reputation for thoroughness and professionalism in his reports, research and teaching. I am sure these characteristics will be evident in this evenings presentation, and with this in mind it is with great pleasure and anticipation that I call upon Professor Robert Mair to deliver the 46th Rankine Lecture.

site. He clearly made his mark elsewhere on the Cambridge scene, and it was not long before he came to the attention of the fellows at Jesus College, who elected him Master in 2001. With all these responsibilities, it seems hard to imagine there is any time to relax. In his spare time he will enjoy a game of tennis or a round of golf and go sailing when the opportunity arises. But, more importantly, Robert always makes sure he has time for his familyhis wife Margaret and children Julia and Patrick. Robert is truly one of the good guys. He is always friendly and hospitable, always polite and generous, and always interested in the people he meets. He is invariably calm and relaxed, and puts people at ease. He is exceptionally well organised, a great team leader who naturally gets the best out

 Mair, R. J. (2008). Geotechnique 58, No. 9, 695736 [doi: 10.1680/geot.2008.58.9.695]

Tunnelling and geotechnics: new horizons

R . J. M A I R * New developments in both the theory and the practice of tunnelling are covered in the lecture. The important relationship between tunnelling and geotechnics is highlighted, and recent advances in research and practice are described, drawing on model studies, theoretical developments and eld measurements from case histories from around the world. Simplied plasticity models are presented that can be used by designers to assess ground movements and tunnel lining loads in complex ground conditions. The important role of pilot tunnels and in situ measurements to validate such models, drawing on a case history from Bolu, Turkey, and on other tunnelling projects, is described. Recent technical advances in earth pressure balance tunnelling are considered, illustrated by measurements from the Channel Tunnel Rail Link project, with emphasis on key factors inuencing volume loss, such as face pressure, soil conditioning and effective screw conveyor operation. A recent case history in Bologna is described, in which the innovative use of directional drilling to install curved grout tubes was employed for a compensation grouting project in granular soils. Timedependent ground movements and tunnel lining distortions occurring after tunnelling are discussed, their magnitude depending on the relative permeability of the tunnel lining and soil, the degree of anisotropy of the soil permeability, and the initial pore pressure prior to tunnelling. The effects of tunnelling-induced settlements on pipelines are considered, drawing on centrifuge tests and analytical solutions, and a new design approach is presented, taking into account the reduction of soil stiffness with increasing shear strain as a result of tunnel volume loss. The lecture concludes with a description of a distributed strain sensing technique using bre optic technology, based on Brillouin optical time domain reectometry (BOTDR), and its innovative application to eld monitoring of a masonry tunnel subjected to new tunnel construction beneath it. La communication porte sur des developpements nou veaux dans la theorie et la pratique du percement de tunnels, et met en valeur les rapports existants entre le percement de tunnels et la geotechnique. On y decrit des ` progres realises recemment, decoulant detudes de mod` eles, de developpements theoriques et de mesures sur place issus dhistoires de cas dans le monde entier. Des ` ` modeles a plasticite simpliee, pouvant etre utilises par des concepteurs pour evaluer les mouvements du sol et les charges des revetements du tunnel dans des conditions complexes du sol, sont decrits. On y decrit egalement le role important que jouent des tunnels pilote et des ` mesures in-situ dans la validation de ces modeles, sur la ` base dune histoire de cas a Bolu, en Turquie, et de projets de percement de tunnels divers. On examine des ` progres techniques realises recemment dans le percement de tunnels avec equilibre de la pression terrestre (EPB), illustres par des mesures effectuees dans le cadre de la liaison ferroviaire du Tunnel sous la Manche, en mettant laccent sur des facteurs cle inuant sur des pertes de volume, par exemple la pression sur la face, le conditionnement du sol et le fonctionnement efcace du transpor` teur a vis sans n. On decrit une etude de cas effectuee ` recemment a Bologne, dans laquelle on a employe de facon innovante le percement directionnel pour linstalla tion de tubes de scellement courbes, dans le cadre dun projet de scellement de compensation dans des sols granulaires. On y discute de mouvements du sol et de deformations du revetement des tunnels avec le temps, dont la magnitude est fonction de la permeabilite relative du revetement des tunnels et du sol, du degre danisotro pie dans la permeabilite du sol, et de la pression inter stitielle initiale, prealablement au percement des tunnels. On examine en outre les effets, sur les canalisations, du tassement dus au percement de tunnels, sur la base de tests centrifuges et de solutions analytiques ; on presente une nouvelle methode conceptuelle en tenant compte de la reduction de la rigidite du sol avec laugmentation de la deformation de cisaillement decoulant de la perte du volume du tunnel. La communication se termine par la description dune technique de detection distribuee des deformations, faisant usage de la technologie des bres ` optiques, basee sur la reectometrie optique a domaine temporel de Brillouin (BOTDR), et son application inno vante dans les controles in situ sur un tunnel de maconn erie, sous lequel on construit un nouveau tunnel.

KEYWORDS: case histories; centrifuge modelling; design; eld instrumentation; ground movement; grouting; monitoring; numerical modelling and analysis; pipelines; theoretical analysis; tunnels

INTRODUCTION This lecture focuses on a number of new developments in both the theory and practice of tunnellingand on the fundamental role of geotechnics in all of these. The lecture draws on new researchinvolving analysis, centrifuge model tests and eld measurementstogether with a selected number of recent case histories of tunnel construction where there has been signicant innovation.
Discussion on this paper closes on 1 May 2009, for further details see p. ii. * University of Cambridge.

The following ve topics are covered: (a) the role of simplied models and their application to deep tunnels in clays (b) ground movement control (i) advances in earth pressure balance (EPB) tunnelling machine technology (ii) recent developments in compensation grouting (c) long-term ground movements (d ) effects of tunnelling on buried pipes (e) advances in bre optic technology for eld monitoring. These particular topics have been selected because they all 695

696

MAIR and in an axisymmetric sense further back from the heading. This assumption is therefore strictly applicable only to deep tunnels (typically with cover-to-diameter ratios in excess of 5). The tunnel is of outside diameter D, and the lining is installed at a distance P behind the face. It is assumed that the rate of advance of the tunnel is sufciently fast that the clay behaviour around the heading is undrained. There is a build-up of pressure L on the lining, from zero (when the lining is installed) to a maximum value L at some distance back from the face (typically about 2D). As shown in Fig. 1, radial ground movement () at the position of the tunnel extrados begins some distance ahead of the tunnel face, and increases to a value 1 at the point when the lining is installed. Pressure then builds up on the lining as the tunnel face moves away from it, and further radial movement of the soil and lining, 2 , takes place. In the case of open-face tunnelling, particularly where the lining is being installed reasonably close to the face (i.e. for small P/D), the ground response ahead of the tunnel can be idealised in terms of spherical cavity unloading (Mair et al., 1992a; Mair & Taylor, 1993); the inner radius of the sphere is equal to that of the tunnel. This is illustrated in Fig. 2(a). Assuming that the radius of the boundary of the clay is large in comparison with the radius of the tunnel, the soil movement ahead of the tunnel face (at a radius r) is given by classical spherical cavity contraction theory as   su a 2 exp 0:75N 1 (1) a Eu r where a is the tunnel radius, su is the undrained shear strength of the clay, Eu is the undrained Youngs modulus of the clay, and N 0 /su (0 being the total overburden pressure at the tunnel axis). The ground movement at the face, 1 , is given by substituting r a in equation (1) to give su exp 0:75N 1 (1a) 1 a Eu A further idealisation is that, as tunnel construction progresses and the tunnel face moves away from the lining that

have important new implications for the design and construction of tunnels. Also, they all illustrate the fundamental role of geotechnics in enhancing innovation in tunnelling practice. First, the lecture discusses the role of simplied models and their usefulness in application to the design of deep tunnels in clays: this will be illustrated by a recent case history in complex and challenging ground conditions. Second, the lecture focuses on ground movement control: this is undoubtedly a crucial issue for all tunnelling projects in soft ground in urban areas. This topic is divided into two parts: recent advances in earth pressure balance tunnelling machine technology for ground movement control are discussed, and then some important new developments in the practice of compensation grouting are presented. Both of these are illustrated by case histories. The third topic also covers ground movements, but focuses on long-term ground movements and their inuence on tunnel lining behaviour: this is becoming increasingly important, as it is recognised that in some cases ground movement caused by tunnelling can be very signicant, and can continue for many years. Ground movements due to tunnelling, and their effects on structures, are increasingly important as more underground construction is undertaken in urban areas, but their effects on services are all too often neglected. The fourth part of the lecture presents some new insights into the effects of tunnelling on buried pipes. Field monitoring is vital for all tunnelling projects, and the nal part of the lecture addresses some recent advances in bre optic technology for eld monitoring, illustrated by some case histories.

SIMPLIFIED MODEL FOR DEEP TUNNEL CONSTRUCTION IN CLAYS Cavity contraction The behaviour of an advancing tunnel in clay is illustrated in Fig. 1. As a simplication, axisymmetric conditions are assumed: that is, all ground movements around the tunnel are radial, and equal at any radius, so that vertical movements equal horizontal movements. The movements are radial in a spherically symmetric sense at the tunnel heading,

Spherical symmetry: movements radial towards heading

Lining

Cylindrical symmetry: movements radial to lining

Radial ground movement at tunnel extrados 2D

2 1 Distance from tunnel face (tunnel diameters) D 0 Lining installed D 2D

Pressure on lining, L

Li Distance from tunnel face (tunnel diameters) 0 D 2D

Fig. 1. Simplied model for an advancing tunnel in clay

TUNNELLING AND GEOTECHNICS: NEW HORIZONS

X

697

1 2 1
D 2a

X (a) (b)

Fig. 2. Simplied assumptions for a tunnel heading: (a) spherical cavity unloading at tunnel face; (b) section XX, cylindrical cavity unloading away from tunnel face

has been installed, the conditions become more like those corresponding to cylindrical cavity unloading, as illustrated in Fig. 2(b). The radial ground movement r (at a radius r) is given by classical cylindrical cavity contraction theory (Mair & Taylor, 1993) as   su a exp N 1 (2) r 3a 2Eu r where N is the stability ratio (0 L) /su: The immediate lining pressure Li can be derived by considering the elastic-plastic ground response to unloading of a cylindrical cavity, taking into account the stiffness of the lining, as shown in Fig. 3. Line ABD is the calculated response of the ground to cavity unloading, and line XC is the lining response. This classical calculation, linking ground

response to support reaction, is well known in the context of underground support for tunnels in rocks (e.g. Ward, 1978; Hoek & Brown, 1980; Panet & Guenot, 1982). As the total radial stress (acting at the external radius of the tunnel lining) r is released from its initial value 0 at point A, radial soil deformation takes place. Idealising the clay as linear elastic-perfectly plastic, the behaviour is initially elastic until point B, when a plastic zone begins to develop around the tunnel boundary. For N < 1 the cylindrical cavity is elastic. By substituting N 1 (and, associated with this, su 0 L ) into equation (2) it can be easily shown that the straight line AB in Fig. 3, for r a and L r , is given by r 0 2Eu 3a (3)

Elastic response (N equation (3)

1)

Total radial stress, r

Li

Elastic-plastic response (N equation (4)

1)

Slope K (lining stiffness)

X
Initial movement 1 at heading prior to lining installation

1 2

Total radial movement,

Fig. 3. Ground reaction curve (cylindrical cavity unloading): elastic response (equation (3)); elasto-plastic response (equation (4))

698

MAIR

Total radial stress, r

The elastic-plastic ground response, for N > 1, corresponding to the curve BD, can be obtained by rearranging equation (2), putting r a to give   2Eu r 0 su ln (4) su 3asu The ground response ABD in Fig. 3 is often referred to by tunnel lining designers as the ground reaction curve. The tunnel lining is installed at point X, after the radial movement 1 has already occurred, as shown in Fig. 1. Pressure then builds up on the tunnel lining, and the subsequent soil and lining displacement 2 is determined by the stiffness of the lining. Equilibrium is reached at C, and the maximum value of lining pressure (in the short term, under undrained conditions) is Li . The initial ground movement 1 prior to lining installation is of key importance. The effect of a larger value of 1 would be a smaller lining stress Li being obtained. This important effect has been understood for a long time in the tunnelling industry; it is also one of the underlying principles of the New Austrian Tunnelling Method (NATM), in which delay of installation of tunnel support leads to reduced pressures and loads being induced on the support. Hence, if 1 can be predicted, reasonable estimates of the lining pressure can be made. Inuence of lining stiffness The inuence of lining stiffness on the predicted lining pressure can be seen from Fig. 4. A more exible lining results in a lower pressure, as can be seen from the line XC1 compared with XC. The radial stress on the lining is given by Li K2 (5)

0 A

C2 Li C C1

Very stiff X 1 Very flexible Total radial ground movement,

Fig. 4. Inuence of lining stiffness on predicted lining pressure: very stiff lining gives upper bound to lining pressure Li ( 1 )

where K is an equivalent spring stiffness relating the inward movement of the lining, 2, to the radial stress acting on it. For a solid tunnel lining of thickness t and outer diameter D, and assuming that t/D is small, it can be shown (Ward, 1978) that K% 4El t D D t (6)

where El is the Youngs modulus of the lining. A generalised form of the simplied model for tunnel construction in clay soils is shown in non-dimensional form in Fig. 5. Different ground response curves, derived from equations (3) and (4), are shown for a range of stability ratios (Mair et al., 1992a). The initial ground movement 1 at the tunnel face is assumed to be that calculated from the spherical cavity contraction idealisation (equation (1a)). The

r/su

N*

Slope Ka/Eu 2

N*

N*

4 N* 5

0 0 20 (/a)(Eu/su) 40 60

Fig. 5. Normalised ground reaction curves for design: N (Mair et al., 1992a)

0 /su ; K

lining stiffness

TUNNELLING AND GEOTECHNICS: NEW HORIZONS lines representing the lining response, with slope Ka/Eu , are shown with a slope of 20, which is typical for many tunnel linings in stiff clays. The intersection of the ground response curves with the lines representing the lining response enables designers to rapidly assess the immediate lining pressure (expressed in terms of the total radial stress r normalised by the undrained shear strength su ). Upper bound to lining pressure: a very stiff lining A very stiff lining is represented by the vertical line XC2 in Fig. 4 and results in an upper bound to the lining pressure. In this case, the radial soil movement r at r a (given by equation (2)) is equal to the soil movement at the face, 1 (given by equation (1a)). Setting (for r a) equal to 1 and combining equations (1a) and (2) leads to an expression for the tunnel lining pressure as a proportion of the total overburden pressure:     Li :25 1 ln 3 (7) 0 2 0 N Equation (7) is applicable only where both the ground adjacent to the spherical cavity (ahead of the tunnel) and the ground adjacent to the cylindrical cavity (the lined tunnel) are in a plastic state. The ground surrounding the cylindrical cavity will be in an elastic state for N < 1; substituting N 1 in equation (2), setting r 1 (both for r a) and combining equations (1a) and (2), it is found that N 1.87. Hence equation (7) is applicable only for N > 1.87. The ground adjacent to the spherical cavity will be in an elastic state for N < 4/3, and therefore for 4/3 < N < 1.87 the spherical cavity will be in a plastic state with the cylindrical cavity remaining in an elastic state. Hence for 4/3 < N < 1.87 the tunnel lining pressure as a proportion of the total overburden pressure can be derived for a very stiff lining by combining equations (1a) and (3), leading to   Li 2 exp 0:75N 1 (8) 1 0 3N For N < 4/3, ground around both the spherical and cylindrical cavities will be elastic. Substituting N 0 /su 4/3 in equation (1a), the deformation of the elastic spherical contracting cavity is given by 1 3 0 a 4Eu (9)

699

as a proportion of total overburden pressure can be derived for a very stiff lining by combining equations (3) and (9) and putting r Li , leading to the simple relation Li 0:5 (for N < 4) 3 0 (10)

Equations (7), (8) and (10) are shown plotted in Fig. 6, from which it can be seen that the tunnel lining pressure varies between 30% and 50% of the total overburden pressure for the wide range of stability ratio N of 06, the proportion reducing with increasing N. It is interesting to note that the tunnel lining pressure for a very stiff lining, given by equations (7), (8) and (10), is independent of the ground stiffness, and is simply a function of stability ratio N (for N . 4/3). This is because the term Eu /su cancels out in the algebra leading to equations (7), (8) and (10). It is also of interest to note that for weaker ground (i.e. for high values of N ) the ratio Li /0 predicted by equations (7) and (8) reduces (albeit slowly, as shown in Fig. 6). This is because the weaker ground leads to higher deformations occurring ahead of the face prior to installation of the lining; the consequence of more ground deformation before installation is a smaller pressure induced on the lining (see Fig. 3).

Summary The simplied model for characterising ground movements around a tunnel and the development of the immediate short-term load on the tunnel lining can be summarised as follows. First, using the spherical cavity contraction idealisation, estimate the maximum soil movement 1 at the tunnel face prior to installing the tunnel lining. Second, combine this with the cylindrical contraction idealisation and the appropriate lining stiffness to calculate the immediate lining pressure. By assuming that the lining circumferential stiffness is high, simple upper bounds to the lining pressure are derived, giving Li /0 0.30.5 for a wide range of stability ratios.

Hence for fully elastic conditions the tunnel lining pressure

06

COMPARISON OF SIMPLIFIED MODEL WITH FIELD DATA Ground movements The idealised spherical contraction model for the ground movements ahead of the tunnel face (equation (1)) implies a linear relationship between the non-dimensional quantities /a and (a/r)2 ; similarly the idealised cylindrical contraction model (equation (2)) implies a linear relationship between
1)

Li /0

05

Li /0

2/3N * exp(075N *

05 Elastic 04 N* 133 N* 187 Elastic-plastic

Li /0

025

1/N * ln 3/2

Li /0

03

02

01

0 0 1 2 3 N* 0 /su 4 5 6 7

Fig. 6. Short-term lining pressure predicted for very stiff lining

700

MAIR and assuming that N is generally around 2.5 for London Clay, the slope of the line in Fig. 8(b) gives a ratio Eu /su of 300: this is also consistent with analysis of radial soil movements around tunnels in London Clay (Mair & Taylor, 1993). Field measurements of axial ground movements were also reported by Wong et al. (1999), in this case for the Tartaguille tunnel for a TGV project in France. A 14.8 m diameter tunnel in a mudstone of undrained shear strength 1.2 MPa was constructed with an open face, and lined with sprayed concrete, at a depth of 110 m: the range of axial ground movement measurements is shown in Fig. 9. These measurements were obtained by means of horizontal multiple-point extensometers installed in the centre of the tunnel face at two separate locations. Fibreglass face bolts were employed, but Wong et al. concluded that their stiffness relative to the ground was relatively low, and therefore their contribution in reducing the axial ground movements was small. The measurements in Fig. 9 are plotted in nondimensional terms in Fig. 10; as for the London Clay data considered earlier, it has been assumed that the centre of the contracting sphere is one tunnel radius from the tunnel face. The general linearity of the plot in Fig. 10 again conrms that the idealisation of the spherical contraction model is reasonable. The solid line shown in Fig. 10 represents a reasonable t through the data: taking the total overburden pressure of 2.3 MPa assumed by Wong et al. (corresponding to the tunnel depth of 110 m and a bulk unit weight of 21 kN/m3 ), and an undrained shear strength of 1.2 MPa, the ratio Eu /su derived from the slope of the line is 390, giving Eu 469 MPa. Dilatometer tests, self-boring pressuremeter tests and plate load tests were undertaken for the Tartaguille project, from which Wong et al. report that design parameters of su 1.2 MPa and Eu 400 MPa were selected. Taking these parameters, together with the total overburden pressure of 2.3 MPa assumed by Wong et al. (corresponding to the tunnel depth of 110 m and a bulk unit weight of 21 kN/m3 ), the predicted axial ground movement from the spherical contraction model (equation (1)) is shown plotted in Fig. 10. Reasonably good agreement is obtained between the prediction using the assumed design parameters and the eld measurements, conrming the value of the simplied model in validating the selected design parameters. Lining pressures Measurements of lining performance were made for a 4.7 m OD tunnel constructed at a depth of 223 m in hard Boom Clay at Mol, Belgium (Neerdael & de Bruyn, 1989). Full details of the measurements and their interpretation are given by Mair et al. (1992a) and Mair (1993). Fig. 11 shows data from load cells incorporated in the lining, converted to equivalent radial pressure acting on the lining. Construction progress was slow: the time to construct a length of tunnel equal to an excavated diameter was about 20 days. The measurements show a build-up of lining pressure to an approximately constant value after about 60 days, which is equivalent to a tunnel length of about three diameters. The lining pressure predicted by the simplied model is shown in Fig. 11: this assumed a linear elastic-perfectly plastic total stress model (su 1.0 MPa, Eu 400 MPa derived from laboratory and in situ testing), and took into account the actual stiffness of the lining, allowing for wood packing between the lining blocks (Mair et al., 1992a). A lining pressure similar to that predicted by the simplied model was predicted by axisymmetric nite element analyses, assuming various soil models, reported by Mair et al. (1992a) and by Gaerber (2003).

r /a and a/r. These are shown in Figs 7(a) and 7(b) respectively (Mair & Taylor, 1993). Field measurements by Ward (1969) of axial ground movements ahead of an advancing tunnel in London Clay are shown in Fig. 8(a): these are shown plotted in nondimensional terms (see Fig. 7(a)) in Fig. 8(b), assuming that the centre of the contracting sphere is one tunnel radius from the tunnel face (Mair & Taylor, 1993). The tunnel was 4.1 m in diameter at a depth of 24 m, and the measurements were made by means of rod extensometers installed from another tunnel (see Fig. 8(a)). The linear nature of the dimensionless plot in Fig. 8(b) conrms that the idealisation of the spherical contraction model is reasonable. Adopting a linear elastic-perfectly plastic model for the soil behaviour,

2a

N*

/a

(a/r )2 (a)

2a

N*

/a

a/r (b)

Fig. 7. Radial deformation associated with cavity unloading: (a) spherical (tunnel heading); (b) cylindrical (away from tunnel heading)

TUNNELLING AND GEOTECHNICS: NEW HORIZONS

Chamber Tunnel C A B 0010 1525
Axial convergence: mm

701

Chamber

Tunnel r

06 0008
Axial convergence: in

Shield 0006 A Curve A 0004

1016

04

/a

508

02

Linear fit to data

0002 C 5 457 152 (a) Distance ahead of hood: ft 0 Distance ahead of hood: m 0 02 04 (a/r )2 (b) 06 08 10

25 762

Fig. 8. Deformations in front of an advancing tunnel heading in London Clay: (a) after Ward (1969); (b) from Mair & Taylor (1993)
Multiple-point extensometer 5000 Tunnel axis depth 110 m 148 m 4000 Overburden pressure Tunnel diameter 47 m Tunnel depth 223 m Boom Clay, su 10 MPa Predicted by simplified model (30% of overburden pressure) Load cell data from three rings

40

Axial movement: mm

30 20 10 0 0 5

36 m

Lining pressure: kPa

20

3000

2000

1000 10 15 Distance ahead of face: m

0 0 20 40 60 Time: days 80 100

Fig. 9. Field data from Tartaguille tunnel: axial ground movements (Wong et al., 1999)
Predicted from assumed design parameters: su 12 MPa su 12 MPa N * 193 Eu 469 MPa Eu /su 390 N* Eu 193 400 MPa

Fig. 11. Tunnel lining pressures at Mol (Mair et al., 1992a)

0005 0004 0003

0002 0001 0 0 02 04 06 08 10

(a/r )2

Fig. 10. Non-dimensional plot of Tartaguille tunnel axial movements (see Fig. 9)

predicted by the model are in good agreement with eld measurements. The model predicts short-term lining pressure 3050% of overburden for very stiff liningsand this depends only on N ( 0 /su ). The attraction of the model is its simplicity. In this age of increasing availability of sophisticated software for numerical analysis it is all too easy to lose the basic understanding of the essence of the problem, particularly if the ground conditions are complex. Engineers need pragmatic solutions for design that capture the key aspects of ground behaviour, without necessarily reproducing every detail: this was also emphasised by Poulos et al. (2005). The simplied model presented here provides such a pragmatic solution for tunnel construction in clay soils. APPLICATION OF SIMPLIFIED MODEL TO TUNNELLING IN COMPLEX GROUND CONDITIONS: A CASE HISTORY IN BOLU, TURKEY The simplied model described above was successfully applied to evaluate the ground behaviour during recent construction of tunnels in Bolu, Turkey, in complex ground conditions. Problems were encountered during construction of the twin highway tunnels, of about 16 m excavated diameter, in a faulted and highly tectonised sequence of rocks, using sprayed concrete and cast in situ concrete linings. Full details are given by Menkiti et al. (2001a). The

Despite the signicant scatter in the measurements, Fig. 11 shows that there is reasonable agreement between the eld data and the lining pressure predicted by the simplied model. Summary The essence of the simplied model is the idealisation of open-face tunnelling as spherical cavity contraction (at the tunnel heading) and cylindrical cavity contraction. Axial ground movement patterns and short-term lining pressures

/a

702

MAIR
Rockbolts Forepoles, 12 m @ 04 m Inner lining 04 m07 m

Bolu tunnels, which are 3.3 km long, are located in rugged, heavily forested and mountainous terrain. They are also located in a rst-degree seismic environment, close to the North Anatolian Fault Zone, and were signicantly affected by the 1999 Turkish earthquakes; the relevant seismic aspects are considered elsewhere (e.g. Menkiti et al., 2001b; ORourke et al., 2001). Here the focus is on design, construction and tunnel performance under static loading. The tunnel depth reached a maximum of 250 m, with the majority of the tunnels at a depth of 100150 m; Fig. 12 shows 2 km of the southern section of the tunnels. The geology consists of highly tectonised and intermixed series of mudstones, siltstones and limestones, with very stiff to hard, heavily slickensided, highly plastic, fault gouge clay. Generally the ground comprises sub-angular blocks of hard material within a clayey matrix. The proportion of the clayey matrix varies substantially, and the poorest ground comprises zones of uniform fault gouge clay; the locations of some of the more extensive zones are shown in Fig. 12. Groundwater levels were also high, being 4585% of the overburden cover above the tunnels, as indicated in Fig. 12. The combination of poor ground conditions, high overburden pressures and high water levels resulted in extremely difcult tunnelling conditions. The original design was based on the New Austrian Tunnelling Method (NATM), with a sprayed concrete primary lining augmented with rock bolts and light steel ribs. In line with NATM philosophy the primary lining had been intended to support the immediate ground loads, with the aim of providing a relatively stable environment within which the inner lining could be cast to achieve the required long-term safety. The original design cross-section is shown in Fig. 13: the thickness of the shotcrete lining was 450 mm, and the shotcrete design cube strength was 20 MPa. In the more clayey ground, large deformations in excess of 1 m were recorded, as shown in Fig. 14; these were accompanied by serious shotcrete damage and deterioration, in the form of compression crushing and onion peeling. At one location, where the tunnel was being excavated through very poor fault gouge clay, partial collapse of the top heading occurred during bench and invert excavation, as shown in Fig. 15: a bearing capacity failure resulted in excessive settlement of the top heading and failure of the temporary top heading invert. A full design review was undertaken. Information on the geotechnical properties of the ground was very sparse, mainly because of the difculties of undertaking sampling from boreholes in excess of 150 m deep in mountainous terrain. An exploratory pilot tunnel was therefore constructed over a length of 829 m (in the worst ground conditions), with the principal aims being
Extensive fault gouge clay zone

Shotcrete, fcu 045 m thick

20 MPa

0 Scale

5m

Excavated diameter 16 m

Face bolts, 15 m Temporary invert

Fig. 13. Typical cross-section of Bolu tunnels: original design (Menkiti et al., 2001a)
Distance behind face: tunnel diameters 05 10 15 20 25 30 35

40 Reprofiling and repair works

Movement: mm

Time-dependent movements (no tunnel advance) 500 2 1 3 5 4 1000 7% strain 1500

Average settlement of points 1, 2 and 3 Convergence of points 45 (compression positive)

Fig. 14. Bolu tunnels: typical deformations, original design, high-plasticity Flyschoid Clay (Menkiti et al., 2001a)

(a) to dene the geology in advance of the main tunnel drives (b) to take high-quality block samples for laboratory testing (classication, index and strength tests) (c) to measure axial ground movements ahead of the tunnel face, and to undertake in situ tests (pressuremeter tests and pore pressure measurements). The pilot tunnel was circular, and of excavated diameter 5.6 m. The simplied model described earlier was applied to select a suitable thickness of lining for the pilot tunnel. At the stage of designing the pilot tunnel the geotechnical properties of the clayey ground were not well established (particularly its strength), but, as shown in Fig. 5, the stress on the lining predicted by the simplied model is not
Water level (open borehole) Water level (piezometer)

1100

R L: m

Tunnel 900 150 m

700 62 500

63 000

63 500 Tunnel chainage: m

64 000

64 500

Fig. 12. Longitudinal section through southern section of Bolu tunnels (Menkiti et al., 2001a). Ground conditions comprise highly tectonised, intermixed mudstones, siltstones and limestones. The materials exist as blocks of hard material within a very stiff to hard, slickensided, high-plasticity clay matrix (Flyschoid Clay). Fault gouge clay zones comprise very stiff to hard, heavily slickensided, high-plasticity fault gouge clay

TUNNELLING AND GEOTECHNICS: NEW HORIZONS

703

This provided an opportunity to back-calculate the ground loading acting on the pilot tunnel using the classical equilibrium condition for a smooth circular ring illustrated in Fig. 16, assuming axisymmetric conditions and ignoring any bending. The radial ground loading Li is given by Li 2 c t D (11)

Top heading settlement

Failure of temporary top heading invert

Top heading settlement

Fig. 15. Bolu tunnels: top heading collapse mechanism in fault gouge clay (Schubert et al., 1997)

particularly sensitive to the value of N. The extreme range of N was estimated to be 36, for which equation (7) predicts the ratio Li /0 to be in the range 3238.5%. On the basis of this simplied model, the lining stress was estimated to be 35% of the total overburden stress, and consequently a lining thickness of 400500 mm was selected for the pilot tunnel, depending on the ground conditions (the design cube strength for the sprayed concrete was 20 MPa, which is somewhat lower than usual). The pilot tunnel was excavated full face in lengths of 11.5 m at an average progress rate of 1.92.3 m/day. Radial stress cells were incorporated in the shotcrete lining of the pilot tunnel, but these are notoriously difcult to interpret (e.g. Clayton et al., 2002), and therefore reliable measurements of the ground loading on the lining were not obtained. However, at two locations there was a failure of the pilot tunnel invert, one of which is illustrated in Fig. 16.

Li D

ct ct

Actual lining thickness at invert 025 m (design 05 m)

Fig. 16. Bolu pilot tunnels: failure of pilot tunnel in fault gouge clay due to reduced thickness of lining in the invert. Assuming axisymmetric conditions, Li (2c t )/D, where Li total radial stress, c compressive stress in lining, t thickness of lining. (Using this, back-analysis of failure gives Li 40% of the overburden pressure. For comparison, the simplied model gives Li 35% of the overburden pressure.)

where c is the average compressive stress in the lining, and t and D are the thickness and diameter of the lining respectively. Investigation showed that the invert thickness in the region of both failures was only 250 mm, compared with the 500 mm that had been designed. At the time of the invert failure the shotcrete lining was only 14 days old, and in situ cores gave a cube strength of 22 MPa. By applying the commonly assumed factor of 0.67 to the cube strength to obtain the equivalent cylindrical compressive strength and putting this equal to c in equation (11), and by assuming t 0.25 m and D 5.6 m, a radial ground loading acting on the lining of 40% of the overburden pressure is obtained (the pilot tunnel axis was at a depth of 150 m and the bulk unit weight was assumed to be 21.5 kN/m3 ). This compares reasonably well with the value of 35% estimated from the simplied model. Table 1 summarises the geotechnical description and index properties of the various units encountered in the pilot tunnel, and Table 2 lists their measured stiffness and strength parameters. Full details of the laboratory testing are given by Menkiti et al. (2001a). The strength parameters were obtained from shear box tests and triaxial tests on block samples taken from the face of the pilot tunnel. The three geotechnical units of particular signicance were the High PI Flyschoid Clay, the Blocky Flyschoid Clay, and the Area 3 Fault Gouge Clay. Typical results from shear box tests on the High PI Flyschoid are shown in Fig. 17. The shear box tests on the more tectonised samples, which are more representative of the mass strength of the ground, show = n values in the range 0.160.21, 9 corresponding to residual angles of friction of 9128. By assuming a realistic range for the groundwater levels (from the piezometer readings) and applying the effective stress parameters derived from the laboratory tests, estimates of the undrained shear strength of the Flyschoid Clays are in the range 600750 kPa. This is consistent with results of undrained triaxial tests performed on tectonised specimens cut from the block samples and with pressuremeter tests performed in situ in the pilot tunnel. The tectonised Flyschoid Clays illustrated ductile behaviour, as can be seen from Fig. 17. Axial ground movements ahead of the face of the pilot tunnel were measured with the system shown in Fig. 18. A hollow closed-end tube with two sets of sliding joints was grouted into a horizontal borehole, using a high-quality grout at the end of the tube and a weak grout along its length. A measuring rod was inserted to the end of the tube to measure its position relative to a reference plumb line xed to the tunnel lining. As the pilot tunnel face was advanced, the tube was progressively cut off. Measurements using this system were undertaken at three locations approximately equally spaced over a distance of 170 m in the pilot tunnel. Test 1 was undertaken in mixed face conditions, comprising the High PI Flyschoid Clay and the Blocky Flyschoid Clay; Tests 2 and 3 were principally in the High PI Flyschoid Clay. The measured axial ground movements are shown in Fig. 19, and are plotted in dimensionless form (as in Fig. 7(a)) in Fig. 20. In most cases it was not possible to obtain the measurements of the axial ground movement close to the tunnel face, because of interference by the

704

MAIR

Table 1. Geotechnical description and index properties of ground conditions encountered at Bolu (Menkiti et al., 2001a) Unit High PI Flyschoid Clay Blocky Flyschoid Clay Area 3 fault gouge clay (Ch 6414064200) AS/EL fault gouge clays (Ch 6284062905) Metasediments Crushed MCB Sound MCB Consistency Stiff, highly plastic, heavily slickensided, clay matrix with occasional rock fragments Medium plastic, silty clay matrix with gravel, cobbles and boulder-sized inclusions Highly plastic, heavily slickensided, stiff clay gouge Very heavily slickensided, highly plastic, stiff to hard clay fault gouge Gravel, cobble- and boulder-sized shear bodies in soil matrix. Soil matrix is 2060% by volume Crushed, weathered, highly sheared, clayey, very weak rock with slickensided, sandy silty clay fault gouge matrix Fractured but competent rock UCS 612 MPa PI: % 55 25 55 4060 1015 15 NA CP and mineralogy 3550%; smectite, with traces of kaolin 3050%; smectite, with traces of kaolin 3060% 2050%; smectite 525%; illite (58%) and smectite (23%) predominant 020% NA

PI, plasticity index; CP, clay percentage by weight (i.e. ner than 0.002 mm); MCB, metacrystalline basement rock; UCS, unconned compression strength; NA, not applicable.

Table 2. Measured stiffness and strength parameters at Bolu (Menkiti et al., 2001a) Unit Peak strength 9: degrees High PI Flyschoid Clay Blocky Flyschoid Clay Area 3 fault gouge clay (Ch 6414064200) AS/EL fault gouge clays (Ch 6284062905) Metasediments Crushed MCB2 Sound MCB2 1517 2025 1316 1824 2530 2025 55 c9: kPa 100 100 100 100 50 50 1500 Residual strength 9: degrees 912 1317 912 612 2025 1520 NA c9: kPa 50 50 50 50 25 25 NA 500 650 700 NA 825 950 High G0 = v 9

From high-quality pressuremeter tests. MCB, metacrystalline basement rock; NA, not available; PI, plasticity index; 9, effective stress friction angle; c9, effective cohesion; G0 , maximum shear modulus; v , initial vertical effective stress 9

06 05

Sample through intact material Two samples of tectonised material (more representative of mass strength)

04 r 03 02 01 0 912

10

20

30 40 50 60 Horizontal strain: %

70

80

90

Fig. 17. Bolu tunnels: drained shear box tests on tectonised Flyschoid Clay

tunnel construction operations. Nevertheless, there is reasonable agreement for the three different tests, especially considering the complexity of the ground conditions. The general linearity of the data plotted in Fig. 20 conrms the validity of the spherical cavity contraction model, which is the framework for the simplied model. Taking the slope of the plot in Fig. 20 and applying the simplied model, the corresponding Eu /su ratios for the range of su values of 600750 kPa are in the range 80170.

The nal design of the main tunnels proceeded using these geotechnical parameters derived from the pilot tunnel. Various tunnel cross-sections were designed, depending on the ground conditions established by the pilot tunnel: full details are given by Menkiti et al. (2001a). In the worst ground conditions, where there were zones of thick Fault Gouge Clay, it was concluded that top heading stability could not be achieved even with a temporary invert. This was because the depth and size of the tunnel, combined with the low undrained shear strength of the ground (600 750 kPa), meant that a bearing capacity failure was likely beneath the footing of the temporary invert (see Fig. 15). To overcome this problem, as shown in Fig. 21, the contractor chose to drive two pilot tunnels in the bench area, which were then backlled with concrete, thus providing foundations with adequate bearing capacity to support the top heading. Primary support consisted of 300 mm of sprayed concrete, augmented by an 800 mm thick precast lining installed and grouted in place a short distance behind the face (816 m). The bench and deep monolithic invert were installed 2235 m behind the face to achieve ring closure. No rock bolts were incorporated, these being judged to be ineffective in the fault gouge clay. A 600 mm thick inner reinforced concrete lining was nally installed, which had been designed taking into account seismic effects (ORourke et al., 2001). In summary, construction of the pilot tunnel for the Bolu

Stress ratio, / n

TUNNELLING AND GEOTECHNICS: NEW HORIZONS

Hollow 38 mm steel tube with two sets of sliding joints Shotcrete lining Tunnel face Very weak cementbentonite grout

705

Point being monitored

Reference plumb line 56 m

High-strength grout (in a fabric grout bag) Initially 15 m long

Measuring rod

Fig. 18. Bolu pilot tunnel: system for measuring axial ground movements ahead of face

200

150

Test 1 Test 2 Test 3

100

50

4 6 8 10 12 Distance of face from point: m

14

16

Fig. 19. Bolu pilot tunnel: measured axial ground movements ahead of face. Data from three tests over 170 m of tunnel (highPI Flyschoid Clay)

008

006

su 600750 kPa (lab. test data) N * 0/su 40 50 Eu /su 80 170

2a

tunnel project provided an invaluable opportunity to investigate the ground behaviour, take samples, undertake in situ testing, and make measurements of axial ground movements ahead of the tunnel face. The simplied model provided a sound basis for estimating the immediate (short-term) radial stress acting on the pilot tunnel lining. It was then applied to interpret the measured axial ground movements ahead of the pilot tunnel face and establish a consistent set of geotechnical parameters for use in design. It is clear from the pilot tunnel and subsequent investigations that the original design cross-section for the 16 m diameter tunnel, shown in Fig. 13, was failing in compression; the 450 mm thick lining was unable to sustain the lining pressure of around 40% of the overburden pressure indicated by the simplied model (and conrmed by the local failures of the invert of the pilot tunnel). The nal design of the full-size tunnels proceeded successfully using the geotechnical parameters derived from observations made during construction of the pilot tunnel.

Movement towards face: mm

004

002

02

04 (a/r )2

06

08

GROUND MOVEMENT CONTROL: ADVANCES IN EARTH PRESSURE BALANCE (EPB) TUNNELLING MACHINE TECHNOLOGY Ground movements and volume loss A key parameter of major importance in soft ground tunnelling is volume loss. Fig. 22 shows the development of surface settlement as a tunnel progresses (Attewell et al., 1986), and Fig. 23 shows a transverse section through the resulting settlement trough. Extensive eld measurements

Fig. 20. Bolu pilot tunnel: non-dimensional axial ground movements (key as Fig. 19)

/a

x v

Extent of surface settlement trough

Temporary support

Shotcrete lining fcu 30 MPa, 03 m

Excavation area 260 m2 (equivalent diameter 18 m) 0 5m Scale

u s smax y z0 x

Intermediate lining, 08 m Final lining, 06 m Bench pilot tunnel with infill concrete Concrete invert

Fig. 21. Bolu tunnels: nal design for thick fault gouge clay zone (Menkiti et al., 2001a)

Fig. 22. Surface settlement trough above an advancing tunnel (Attewell et al., 1986)

706
y i 2iSmax Vs D2/4 s

MAIR EPB tunnelling: key aspects There have been considerable developments in earth pressure balance (EPB) tunnelling machine technologies in recent years: excellent ground movement control in a wide variety of ground conditions is now achievable, especially in ground that would be unstable in the absence of face support. The essence of an EPB machine is provision of substantial support to the excavated face at all times, thereby controlling ground movements. Fig. 24 shows the principal features of a modern EPB tunnel boring machine (TBM). The primary function of the cutterhead (1) is to excavate the soil; it is powered by the drive motor (2), all of which is within the circular steel skin (or shield) (3) of the TBM. The excavated soil passes into the pressurised head chamber immediately behind the cutterhead. Access into the chamber, if necessary, can be facilitated by means of compressed air being applied and access being via an air lock (4). A key feature of the EPB machine is the extraction of the excavated soil from the pressurised head chamber by means of a screw conveyor (5), which is an Archimedian screw within a cylindrical steel casing. The screw conveyor plays an important role in the excavation process. The soil is extruded along the screw conveyor to the discharge outlet (7), where the soil is discharged at atmospheric pressure onto a conveyor belt (9). The rotational speed of the screw, its geometry, the restriction of the discharge outlet, and the soil properties all inuence the soil ow rate and pressure gradient along the conveyor. The head chamber pressure supporting the tunnel face is regulated by controlling the rate of soil discharge in relation to the advance rate of the machine (and this leads to the pressure dissipation along the screw conveyor). Laboratory tests using an instrumented model screw conveyor with a range of soft clay samples and operating conditions, and their theoretical interpretation, are reported by Merritt (2004) and Merritt & Mair (2006, 2008). The factors inuencing the chamber pressure during the excavation period are complex, but the details of the screw conveyor operation are of particular relevance. It is important that the extraction of the soil is well controlled, synchronised with the speed of excavation, and that the soil mixture is converted to a low-shear-strength paste (typically in the range 2030 kPa) by suitable soil conditioning (Milligan, 2000; Merritt, 2004). Control of soil ow through the screw conveyor is necessary to control the volume of soil discharged, as well as the dissipation of pressure between the head chamber (in which it is high) and the conveyor outlet (which is at atmospheric pressure). If the soil is too uid, control of the ow rate and pressure

Volume Vs Vol. loss Vl

Point of inflection Settlement

smax exp

y2 2i2

Fig. 23. Transverse settlement trough: Gaussian curve

have shown that the settlement trough can be well characterised by the Gaussian distribution (Peck, 1969; Schmidt, 1969; Rankin, 1988; Mair & Taylor, 1997), with the settlement given by the equation   y2 (12) S Smax exp 2 2i The volume of the settlement trough (per metre length of tunnel), VS , can be evaluated by integrating equation (12) to give VS p 2 iSmax (13)

and this volume, expressed as a proportion of the theoretically excavated tunnel volume (usually expressed as a percentage) is the volume loss: VL 4VS D2 (14)

Typical volume losses for open-face tunnelling in soft ground are generally in the range 13% (Mair, 1996; Mair & Taylor, 1997). Closed-face tunnelling, with signicant face support, tends to result in lower values than open-face tunnelling. Open-face tunnelling in London Clay can in some circumstances give rise to higher values: for example, in St Jamess Park in London volume losses as high as 3% were recorded, but this is exceptional, and can be attributed to particular features of the geology and the operational methods of tunnelling (Dimmock & Mair, 2006a, 2006b; Standing & Burland, 2006).

3 6 4

7 2

5 8

Fig. 24. Earth pressure balance tunnelling machine: 1, cutter head; 2, drive motor; 3, TBM skin; 4, airlock; 5, screw conveyor; 6, lining erector arm; 7, soil discharge; 8, lining segments; 9, belt conveyor

TUNNELLING AND GEOTECHNICS: NEW HORIZONS gradient can be problematic, because proper face control requires that the chamber is always lled with soil, whereas if the soil is too stiff the conveyor can require excessive power to operate or it can become jammed. Natural soils do not usually have ideal properties when excavated, and soil conditioning is often used to modify the properties to improve the operation of EPB machines.

707

of the segmental lining and the excavated ground. Tail skin seals prevent the grout from entering the TBM. This process of tail void grouting, together with high-quality face pressure control, is a vital part of ground movement control.

Soil conditioning Soil conditioning is achieved by injecting conditioning agents, most commonly foams or polymers, from the cutterhead and into the head chamber to mix with the soil during the excavation process. The effects of soil conditioning on soil properties are varied and complex: many of them are summarised by Maidl (1995), Leinala et al. (2000), Milligan (2001), Merritt et al. (2003), Merritt (2004), Boone et al. (2005), OCarroll (2005) and Borghi (2006). The properties of the soilchemical mixtures depend strongly on the type and quantity of the conditioning agent, or combination of agents, mixed with the ground. Operation of the tunnelling machine and control of the face pressure may be signicantly affected by these different properties. The parameters that have to be selected for the soil conditioning comprise the type of agent (water, bentonite, polymer, foam or any combination of these) and their quantities. Further details of soil conditioning, denitions of injection parameters and typical quantities used for different ground conditions are given in Appendix 1. Conditioning agents are sometimes also injected into the screw conveyor in order to modify further the properties of the spoil as it passes through the conveyor. This can also have the effect of reducing the screw torque. However, the sensitivity of the screw conveyor operation to the shear strength of the spoil suggests that the excavated soil should best be conditioned as early as possible (i.e. at the cutterhead and in the head chamber before it enters the screw conveyor) in order to maximise the mixing time and hence improve the homogeneity of the spoil in the excavation chamber (Borghi, 2006).

Provision of face pressure In an EPB machine the support pressure to the excavation face is provided partly by the thrust from the cutterhead and partly by the chamber pressure, the relative proportion depending on the opening ratio , dened as the ratio of the total openings surface area A0 to the total face area A. Values of vary for different machines; for larger values of control of face pressure will depend more on the chamber pressure. The pressure in the excavation chamber is controlled by the mass ow rate of soil and conditioning agents entering the chamber, and by that of the spoil discharged at the outlet of the screw conveyor. If the machine advances steadily, a reduction in the screw conveyor extraction rate will cause an increase in pressure in the excavation chamber: correspondingly, an increase in extraction rate will result in a reduction in chamber pressure. The control of the extraction rate is strongly affected by the mechanical properties of the spoil. The bulk modulus of the spoil (i.e. the mixture of soil and conditioning agents) also has a strong inuence on the chamber pressure uctuations. Compressible mixtures with low modulus (e.g. sand and foam) give less uctuations. However, if the excavation chamber is full of spoil and the mixture is almost incompressible (for example if it is principally clayey soils and liquid conditioning agents), differences in the rates of excavation and spoil discharge may result in signicant pressure uctuations (Borghi, 2006).

Tail void grouting Following an excavation cycle, when the jacks are retracted, tunnel lining segments, (8) in Fig. 24, are erected within the TBM tail skin by means of an erector arm (6). As the tail skin leaves the tunnel lining, grout is injected under pressure to ll the annular void between the extrados
West Portal 40 30 20

EPB tunnelling on CTRL Contract 220 A simplied view of the geology for Contract 220 of the Channel Tunnel Rail Link (CTRL) in London is shown in Fig. 25. The tunnelling works on Contract 220 comprised 7.5 km of twin tunnels of outside diameter 8.1 m excavated with EPB machines westwards from the Stratford Box to the portal near St Pancras station in London. Full details of the project are given by Woods et al. (2007). A very wide range of ground conditions was encountered, as can be seen from Fig. 25: details are summarised by Borghi (2006). Upon launching from Stratford Box the tunnelling machines rst encountered about 80 m of mixed face conditions of the

Groundwater table in upper aquifer Stratford Box Made Ground and Terrace Gravel

Elevation: m

10 0 10 20 30 40

London Clay Woolwich and Reading Beds Upnor Formation Thanet Sand

Chalk Groundwater table in lower aquifer 2000 3000 4000 5000 6000 7000 8000 9000

Chainage: m

Fig. 25. Geology of CTRL Contract 220

708

MAIR
300
Excavation Ring build Excavation Ring build Excavation Ring build

Lambeth Group (Upnor Formation (UF), comprising sands, silts, clay, gravels and pebbles, and the Woolwich and Reading Beds, comprising very stiff clays). The proportion of the Lambeth Group soils then gradually reduced for the next 140 m, while the very dense silty Thanet Sand appeared in the invert; full-face Thanet Sand was encountered after 560 m of tunnel drive. Subsequently chalk was encountered in the invert, before the tunnels started rising in elevation crossing the full sequence of tertiary soils, including the gravels of the Harwich Formation (HF) at the base of the London Clay, with full-face London Clay in the nal 1000 m of the drive. The tunnels were thus driven in a very wide variety of ground conditions, varying from dense sands and gravels to stiff clays and chalk, often in mixed face conditions. Research has been undertaken at Cambridge University (Wongsaroj, 2005; Borghi, 2006) in collaboration with the contractor, Nishimatsu Construction, and the client, Rail Link Engineering. Surface settlement measurements from 48 instrumentation arrays were analysed (Wongsaroj et al., 2005), and the results in terms of volume loss measured at the ground surface are shown in Fig. 26. It can be seen that, with only two exceptions, the measured volume loss was always less than 0.8% and often as low as 0.2%. The high degree of ground movement control illustrated in Fig. 26 can be attributed to a number of important advances in EPB technology (British Tunnelling Society, 2005; OCarroll, 2005). There have been signicant developments in the technique of tail void grouting; good control of pressure and volume of grout injection is essential for effective control of ground movements. Also, lling the annulus around the EPB shield with a bentonite paste, as was undertaken on CTRL Contract 220, can signicantly reduce ground movements. Another key aspect of ground movement control is the control of chamber pressure. Fig. 27 shows an example of difcult chamber pressure control during three cycles of excavation and ring build during tunnelling in the Lambeth Group (Borghi & Mair, 2006): this example has been deliberately selected as a case of poor pressure control. The average chamber pressure (recorded by ve pressure sensors) is plotted against time for the three construction cycles, each comprising two phases: the excavation phase, when the EPB machine is advancing; and the ring-build phase, when the machine is stationary during erection of the tunnel lining.
West Portal 14 12

FIR LIR

92% 15%

FIR LIR

84% 14%

FIR LIR

90% 18%

Chamber pressure: kPa

200

100

Average chamber pressure

05

10

15 20 Elapsed time: h

25

30

35

Fig. 27. Example of difcult face pressure control in the Lambeth Group formation (Borghi & Mair, 2006): FIR, foam injection ratio; LIR, liquid injection ratio (see Appendix 1 for denitions)

The foam and liquid injection ratios (see Appendix 1 for the denitions) recorded for these cycles were in the ranges 84 92% and 1418% respectively. The duration of the excavation phase was typically 0.71 hthe time taken for the EPB machine to advance 1.5 m, which is the length of one ring of the tunnel segments. During this excavation period it can be seen that the chamber pressure uctuates signicantly, sometimes rising substantially: there is no clear explanation for this, but it may be a consequence of a reduction in screw conveyor extraction rate, as discussed earlier, as well as injection of signicant quantities of pressurised foam into the excavation chamber. Figure 27 shows that there was always a substantial drop in chamber pressure during the ring-build phase, when the machine was stationary for 0.30.5 h. This may be because the chamber was not completely full of spoil, but it is also believed to be partly due to the foam breaking down during stoppage of the machine owing to sorption of the foaming liquid into the clay (Borghi, 2006). This process of sorption was also observed in the laboratory (Mair et al., 2003; Merritt et al., 2003). The injection of signicant quantities of foam may act counter-productively when tunnelling in
Stratford Box Thanet Sand (TS) TS TS Chalk UF

Volume loss: %

10 08 06 04 02 0 1000

Woolwich and Reading Beds (WRB) WRB WRB WRB LC UF HF HF

London Clay (LC) 3000 5000 Chainage: m 7000 9000

Fig. 26. Observed volume loss for Up-Line tunnel CTRL Contract 220 (Wongsaroj et al., 2005): UF, Upnor Formation; HF, Harwich Formation

TUNNELLING AND GEOTECHNICS: NEW HORIZONS clays by exacerbating the drop of pressure in the head chamber during the ring-build phase; this is particularly the case if this phase takes longer for any reason, as the foam breaks down and the spoil mixture effectively compresses. In contrast, Fig. 28 shows an example of good control of chamber pressure (Borghi, 2006). A number of cycles of excavation and ring build in a full-face of London Clay are shown; the tunnel axis was at a depth of 31 m. In the upper part of the gure, the average measured chamber pressure is plotted against time; in the lower part the total shield thrust (measured from the total load of the jacks reacting off the most recently completed segments) is plotted against time. During the excavation cycles the shield thrust is reasonably constant at around 20 MN. During the ring build, between the excavation cycles, the thrust is considerably reduced as jacks are retracted to allow erection of the lining segments. Fig. 28 shows that there is sometimes a drop in chamber pressure during the ring-build phase (this may be partly a consequence of the corresponding reduction in shield thrust). Two ring-build cycles are highlighted, one showing a larger chamber pressure drop and the other a smaller drop. Nevertheless, the average chamber pressure throughout the period of 8.5 h shown in Fig. 28 remains reasonably constant at around 220 kPa, with uctuations generally not exceeding 30 kPa. The average chamber pressure p of 220 kPa in Fig. 28, expressed as a ratio of the total overburden pressure v0 at tunnel axis level, is p/v0 0.35. It is reasonable to expect this average chamber pressure ratio to be linked to the volume loss, and at rst sight this is indicated by the data in Fig. 29, which show different face pressure ratios plotted against measured volume loss for seven tunnel sections in full-face London Clay (after Wongsaroj et al., 2005). However, it should be noted that six of the seven cases are for tunnels with axis depths of 2530 m, and for these cases there appears to be no signicant reduction of volume loss with increasing average chamber pressure ratio. This can possibly be attributed to the small stability ratio and load factor prevailing in the stiff London Clay at these depths (Mair et al., 1981; Macklin, 1999). It should be noted, however, that the chamber pressure was varied only over a small range for these six cases at tunnel axis depths of 2530 m. Also, the small volume losses measured corresponded to only a few millimetres settlement at the ground surface, and therefore the assessed volume losses are prone to error. It should also be recalled that the chamber pressure is only one component of the total face pressure, albeit the major one for many tunnelling machines for
Chamber pressure: kPa
10 08 06 04 02 0 0 02 04 06 08 Average pressure/v0 10 Axis depth 10 m bgl Axis depth 2530 m bgl

709

Fig. 29 Inuence of EPB chamber pressure on volume loss for full face in London Clay (Wongsaroj, 2005; Borghi, 2006)

which the opening ratio is large (as for the case of the machines on CTRL Contract 220, where was 57%). Nevertheless, Fig. 29 shows that a signicantly lower volume loss was measured when a higher average chamber pressure ratio was operating for the much shallower tunnel in London Clay, for which the axis depth was only 10 m below ground level. The immediate volume loss can be split into different components (Dimmock, 2003; Dimmock & Mair, 2006a). In the case of EPB closed-face tunnelling these were dened by Wongsaroj et al. (2005) as follows: (a) ahead of the face, termed face volume loss (b) around the shield, termed shield volume loss (c) around the tailskin and tunnel lining, termed tailskin volume loss. Shirlaw et al. (2003) advocated that, for soft clays, occasional low face pressures (by which they meant chamber pressures) may have a signicant effect on face volume loss, because small face pressures cause stress relief (and associated ground movements) that cannot easily be reversed by increasing the face pressure. However, Wongsaroj et al. (2005) and Borghi (2006) found that, in the case of the fullface tunnels in London Clay, there was little or no correlation between component 1 and p10% , where p10% is dened as the 10th percentile of the distribution of the chamber pressure p, that is, the value of the chamber pressure below which 10% of the measured values fell. This is probably because such uctuations in chamber pressure for tunnels at great depth in stiff London Clay have only very small effects on the stability ratio and load factor (as discussed earlier for the average chamber pressure ratios shown in Fig. 29). Fluctuations in chamber pressure appear to be of considerably more signicance in the case of shallow tunnels. This is discussed in the following section. Shallow tunnels and piled foundations As part of his PhD studies at Cambridge University, Selemetas (2005) reported measurements of the eld response to tunnel construction of instrumented piles and a number of piled structures on Contract 250 of the CTRL project. Fig. 30 shows a cross-section through the two tunnels at shallow depth beneath part of a 250 m long piled reinforced concrete rectangular culvert. Because of the presence of 5 m of very soft and compressible silts and peat, the culvert was supported by 6 m long piles at 4 m spacing driven into the underlying dense sand and gravel. The 8 m diameter CTRL tunnels were constructed, using EPB Lovat tunnelling machines, with their crowns 4 m below the base of the piles and with only 3 m of cover of London Clay. The up-line tunnel was constructed rst, and caused 3 mm

300 250 200 150 100 50 0 30 25 20 15 10 5 0

Ring build 0 1 2 3 4 5 6 7 8 9

Shield thrust: MN

5 4 6 Elapsed time: h

Fig. 28. EPB chamber pressure and shield thrust for full face in London Clay (Borghi, 2006)

VL: %

710
3m 5m 3m Fill Moggs Farm Culvert Silt and peat Dense sand and gravel London Clay Clay cover 26 m 6m 4m 3m

MAIR vances in soil conditioning technologies, and through a combination of laboratory and eld measurements there is now an improved understanding of the interaction of soilconditioning agents with excavated soils. Model tests have also provided insight into the interaction between conditioned soil mixtures and the operating conditions of screw conveyors. Good control of face pressure, through proper control of the excavation chamber pressure, depends on appropriate soil conditioning and screw conveyor operation. During the excavation phase, the pressure in the chamber is controlled by the mass ow rate of soil and conditioning agents entering the chamber and that of the spoil discharged at the outlet of the screw conveyor. During the ring-build phase, when the tunnelling machine is stationary, there is a tendency for the chamber pressure to drop, especially if the chamber is not completely full of spoil, and this may be exacerbated by longer stoppage times and excessive use of foam for conditioning of clay soils. For EPB tunnelling in stiff London Clay, in cases where the axis depths were 2530 m, there appeared to be no signicant reduction of volume loss with increasing average chamber pressure ratio (at least for the range of pressures measured). At shallow tunnel depth, however, the average chamber pressure ratio was more signicant in reducing volume loss. Chamber pressure uctuations are also of more signicance in relation to ground and structure movements in cases of shallow tunnels.

8m Down-line tunnel 14 mm settlement of culvert VL 05% Up-line tunnel 3 mm heave of culvert

Fig. 30. Effect of EPB tunnelling on piled culvert at Moggs Farm (Selemetas, 2005)

of heave of the culvert above the tunnel centreline. Shortly afterwards the down-line tunnel was constructed, but this time 14 mm of settlement was experienced by the culvert above the tunnel centreline owing to construction of the down-line tunnel alone. An insight into the difference in response can be gained from the chamber pressure records for the two tunnels, shown in Fig. 31. The average chamber pressures were similar: 1.5 bar and 1.3 bar respectively for the up-line and down-line tunnels (1 bar 100 kN/m2 ); but it is evident that there was considerably more uctuation in chamber pressure for the down-line tunnel. Signicantly, the p10% value for the down-line tunnel was 0.9 bar, which was considerably less than the corresponding value of 1.3 bar for the up-line tunnel. The control of chamber pressure was much better for the up-line tunnel, with much less uctuation. Summary Good control of face pressure, through effective control of chamber pressure, is now possible with EPB tunnelling machines in a wide variety of ground conditions. Low volume losses, well below 1%, are now readily achievable. The principal features of successful EPB machine operation are the pressurised excavation chamber, the conditioning of the excavated spoil, the screw conveyor, and both annulus and tail void grouting. There have been considerable adFace pressure: bar
4 3 2 1 0 05 10 15 20 25 Time: days after 15/02/03 (a) Avg 30 35 Avg 152 bar

RECENT DEVELOPMENTS IN COMPENSATION GROUTING: A CASE HISTORY IN BOLOGNA Principles of compensation grouting Compensation grouting is a very promising technique being used increasingly to control ground and building movements during tunnelling in soft ground. The principles of the method were presented by Mair & Hight (1994), and the basic concept is illustrated in Fig. 32. Grout is injected between the tunnel and the building foundations to compensate for ground loss and stress relief caused by the tunnel excavation. Sleeved grout tubes (tubes a manchette, TAMs) ` are installed in the ground prior to tunnelling, often from a vertical shaft. Before tunnelling commences, conditioning grouting is undertaken to tighten the ground and reverse any settlement or loosening of the ground caused by drilling for TAM installation. Grout injection is then undertaken simultaneously with tunnelling in response to detailed observations, the aim being to limit building settlements and distortions to specied, acceptable amounts. Experience of compensation grouting is reported by, among others, Mair et al. (1994), Harris et al. (1994), Harris et al. (1996), Mair & Taylor (1997) and Harris (2001). The technique was successfully used on the Jubilee Line Extension Project in London for the protection of many historic buildings, including the

Face pressure: bar

4 3 2 1 0 05 10 15 20 25 Time: days after 16/03/03 (b)

127 bar Grout injection

Slight settlement with compensation grouting

30

35

Excavated shaft Tube manchette (TAM) Compensation grouting zone Tunnel

Ext

trou of ent

Fig. 31. EPB chamber pressures at Moggs Farm (Selemetas, 2005): (a) up-line tunnel (slight heave); (b) down-line tunnel (signicant settlement)

Severe settlement gh without compensation grouting

Fig. 32. Principle of compensation grouting

TUNNELLING AND GEOTECHNICS: NEW HORIZONS Big Ben Clock Tower at the Palace of Westminster (Harris et al., 1999; Harris, 2001). The majority of experience of compensation grouting has been in stiff clays, and in most cases the grout tubes have been installed from vertical shafts. The recent case history in Bologna, described in the following section, illustrates the success of the technique in granular soils and also introduces the innovative use of directional drilling for installation of the grout tubes.

711

cracking of the masonry arches, some of which were already cracked. Compensation grouting was therefore implemented during tunnelling, with the grout injected in the nominally 4 m thick zone shown in Fig. 33. Also shown in Fig. 33 are the three lines of automatic water level settlement gauges, which provided real-time monitoring data (further details are given by Kummerer et al., 2007); precise levelling was also undertaken. Directional drilling Figure 34 shows a cross-section through the viaduct and the underlying tunnels. The viaduct crossed busy streets, and suitable shaft locations were not available; the contractor therefore proposed that curved TAMs be installed by directional drilling from shallow pits, typically about 2 m deep. Site trials were conducted to prove the feasibility of this. For a 60 m long tube the typical vertical control was up to 1 m, and the horizontal control up to 0.5 m; the minimum radius of curvature was 90 m. Two layers of TAMs were installed, with very tight control on the directional drilling operations, such that the nominally 4 m thick treatment zone was no closer than 1.5 m below the pier foundations and 3 m above the crown of tunnel 2. The directional drilling in progress is shown in Fig. 35. Fig. 36 shows a plan diagrammatic view of the full coverage of TAMs beneath the foundations of the
Temporary pit 601 577 15 m Two layers of curved TAMs 0 10 m Scale Tunnel 1 Tunnel 2 594 Treatment zone Existing NaplesMilan rail link on masonry viaduct

The Bologna project The project was part of a new high-speed rail link under construction between Milan, Rome and Naples. A section in Bologna passes beneath a number of railway bridges and follows the alignment of the existing railway. The most important of these bridges is a nine-arch masonry viaduct, which is 112 m long. Fig. 33 shows a longitudinal section through the viaduct and the new high-speed rail twin tunnels of diameter 9.1 m that were constructed parallel to the viaduct and directly beneath its alignment. The viaduct comprises nine masonry arches, each of 8 m clear span, with one longer 16 m clear span. It is 112 m long and 11 m wide, with 810 m high embankments at each end. The piers of the viaduct are supported on shallow foundations. The ground generally comprises made ground up to 8 m thick over a substantial depth of alluvial deposits, which are predominantly dense gravelly sands or very sandy gravels, with nes content (particles , 0.063 mm diameter) generally less than 15% by weight. Coring through the masonry piers indicated that at least two of the pier foundations extend to the top of the sands and gravels. The crown of the tunnels would be 10.6 m below the deepest pier foundations, that is, a distance of just over one tunnel diameter. The groundwater table was below tunnel invert. Construction of the twin 9.1 m tunnels, using EPB tunnelling machines, was expected to generate large settlements, typically around 20 mm but potentially up to 50 mm for volume losses of 1%. Such settlements would have induced excessive distortions of the viaduct, which was a cause of concern, particularly as suspension of train services was not permitted. There were also major concerns about potential
Three strings of water level settlement gauges

44 m

30 m

420

Fig. 34. Bologna tunnels: typical cross-section showing compensation grouting tubes and treatment zone

Existing NaplesMilan rail link on masonry viaduct

Made Made Ground Silt and clay Sand Ground

179 m
Sand

106 m

Gravel

Gravel

91m

Sand

Gravel

Tunnel As-built extent of treatment zone

Fig. 33. Bologna tunnels: viaduct and geotechnical section

712

MAIR structure to 1 : 1000, and this was adopted as the maximum tolerable value or the alarm value for the viaduct. A trigger level of differential settlement of 1 : 3000 was accordingly established, and the contractor was obliged to carry out compensation grouting if this trigger level was reached. Figure 37 shows the cumulative volume of grout injected during tunnelling, together with the EPB machine progress for the rst tunnel, both plotted against time. The average progress rate for the tunnelling was 23 m/day. The measured volume loss immediately prior to the tunnel reaching the viaduct was 0.2%. Earlier it had been signicantly larger up to 1%but a high standard of face pressure control and tail void grouting was exerted as the tunnelling machine approached the viaduct. Also plotted in Fig. 37 is the ratio of volume of injected grout to volume of ground excavated, expressed as a percentage. During the period when the tunnelling machine was beneath the viaduct (and within 4 m beyond each end), this ratio was in the range 0.40.6%. Grouting continued when the tunnelling machine passed beyond the viaduct, and the ratio correspondingly increased.

Fig. 35. Bologna tunnels: directional drilling in progress to install grouting tubes

Tunnel 2 Tunnel 1

TAMs

Foundations of masonry viaduct 100 m

Drill rig positions

0m

Fig. 36. Bologna tunnels: plan view and layout of grouting tubes

viaduct. A TAM installation progress rate of 65 m/day/rig was achieved. Compensation grouting A eld trial was carried out adjacent to the viaduct, in advance of the main works. The trial was used to rene the grout mix and grouting processes for the ground conditions, and to demonstrate the ability to generate controlled heave. The trial also allowed evaluation of the proposed monitoring system. Some temperature sensitivity was observed under the diurnal temperature variations of up to 308C, measured on the viaduct. Two grouting stages were undertaken prior to commencement of tunnelling and compensation grouting: (a) pre-treatment grouting to permeate and ll the larger voids in the ground; and (b) conditioning grouting to tighten the ground and reverse any settlement or loosening of the ground caused by drilling for TAM installation. A relatively uid grout mix was adopted by the contractor. The water : cement ratio was 0.78; bentonite was added (weight of bentonite powder to weight of water 11%), resulting in a ratio of volume of solids to total volume 23%. However, after injection into the granular soils the process of pressure ltration would rapidly squeeze out the water from the mix, leaving a more viscous mix that would tend to remain relatively local to the injection point. The viaduct was monitored against predetermined, tolerable settlement and differential settlement limits. Compensation grouting activities were implemented as necessary to keep the movements of the structure within these limits. The masonry of the viaduct was already cracked in a number of locations, which were a cause of concern. Following detailed structural analysis of the viaduct, project contract specications limited differential settlement for each span of the

Performance of the viaduct Figure 38 shows the performance of the viaduct at a particular stage of the construction of tunnel 1 (when the face of the TBM was at chainage 3410). Also shown is the estimated transient longitudinal settlement prole without any grouting, assuming a volume loss of 0.2%, and a trough width value K 0.4 (Mair et al., 1996), based on greeneld measurements. The actual settlements of the viaduct achieved at this stage are shown in Fig. 38 (for both the west and east sides of the viaduct); at one location, at chainage 3330, a slight heave was measured. The grout intensity in litres/m2 is also shown, assuming the area of coverage of each grout injection in plan was 3 m 3 3m, based on observations during the eld trial; the inuence of the assumed area of coverage on derived grout intensity is discussed by Viggiani (2001). Slopes corresponding to the trigger and alarm differential settlement levels of 1 : 3000 and 1 : 1000 respectively are shown in Fig. 38, and it can be seen that grouting was necessary because the trigger level was exceeded. Excellent control of the viaduct movements was achieved. It should be recalled that the tunnelling volume loss was only 0.2% while the EPB machine was beneath the viaduct. If the volume loss had been as high as 1%, which had been measured earlier, the likely settlement of the viaduct due to tunnel 1 in the absence of grouting would have been 50 mm rather than 10 mm: signicantly more grouting would then have been necessary to maintain
Grout volume: m3
120 80 40 0 28/6/05 Volume loss 02% 30/6/05 02/7/05 04/7/05 Time 4m 06/7/05 Cumulative grout vol. % Grout vol./Excavated vol. 12 08 04 0

Chainage: m

3500 3400 3300 28/6/05 Viaduct border 30/6/05

02/7/05 04/7/05 Time

06/7/05

Fig. 37. Bologna tunnels: injected grout volume and progress of TBM 1

Grout: % of excavated vol.

TUNNELLING AND GEOTECHNICS: NEW HORIZONS

713

Ave grout intensity: l/m2

100 50 0 3430 2 2 West profile East profile 6 1:3000 Grouting trigger 10 14 TBM1 18 Chainage of TBM1: m 1:1000 Contractual limit No grouting VL 02% 3410 3390 3370 3350 3330 3310 Grout intensity 2/07/05 14:00

Fig. 38. Bologna tunnels: performance during TBM passage. The maximum permitted differential settlement between adjacent piers allowed by the contract was 1 : 1000. The design required compensation grouting injections to be implemented if differential settlement between adjacent piers exceeded 1 : 3000

the differential settlement within the contractual limit of 1 : 1000. Summary The compensation grouting at Bologna has demonstrated the innovative use of directional drilling to install curved grout tubes. This is of practical importance for projects in congested urban areas where it might not be possible to construct shafts. The project has also demonstrated the successful application of compensation grouting to granular soils, for which there has been generally less experience in comparison with clay soils. The importance of eld trials cannot be overemphasised: these were vital to prove the feasibility of both the directional drilling and the proposed grout mixes prior to tunnelling. The sensitive masonry viaduct experienced only small and acceptable levels of distortion, and the existing rail services continued without interruption. The compensation grouting provided a high degree of control. LONG-TERM GROUND MOVEMENTS Inuence of drainage into tunnels Our understanding of tunnelling-induced ground movements and settlements is centred principally around the immediate, short-term movements associated with tunnel construction. However, tunnelling in low-permeability soils often results in ground surface settlement that continuously increases over a long period of time. If the tunnel acts as a drain, it introduces a new drainage boundary condition that leads to long-term reductions in pore pressure and associated consolidation settlements, as depicted in Fig. 39. This is because, on the inside face of the tunnel lining, the pressure is usually atmospheric. If the tunnel is not totally impermeable, a ow of pore water into the tunnel occurs, and a new steady-state ow condition is eventually reached. The nal pore pressures will generally be lower than those prior to tunnel construction: settlement will therefore occur as pore pressures reduce to their long-term steady-state values, increasing effective stresses and thereby inducing consolidation

Heave: mm

Immediate (undrained) settlement Long-term (drained) settlement Tunnel squats

Fig. 39. Tunnel in clay acting as long-term drain

in the clay. As indicated in Fig. 39, the resulting settlement prole at the ground surface will tend to be considerably wider than the prole associated with construction (Mair & Taylor, 1997). There will also be an accompanying tendency for the tunnel to squat with time as consolidation occurs, that is, reduce in vertical diameter and increase in horizontal diameter, as shown in Fig. 39: this has been noted for tunnels in London Clay and in other clays (Ward & Pender, 1981). The evidence that tunnels in low-permeability soil act as new drainage boundaries has been demonstrated from eld measurements of pore pressure around tunnels (Ward & Thomas, 1965; Palmer & Belshaw, 1980). Ward & Pender (1981) concluded that in most cases segmentally lined tunnels in London Clay act as drains, despite the linings having been grouted. This is generally conrmed by recent measurements of pore pressures around ve very old London Underground tunnels in London Clay, shown in Fig. 40. The measurements were all made at tunnel axis level at various distances from the tunnel extrados. Each of these tunnels, which are at least 80 years old, had been constructed with bolted cast iron linings, which were grouted. Details of the year of construction and the depth of each tunnel are given in Table 3. In cases AD, as can be seen from Fig. 40, there is a clear trend of decreasing pore pressure close to the tunnel, and the pattern is reasonably consistent with the pore

714
250

MAIR
Oval Golders Green 1 Golders Green 2 200 Aldwych Kennington 150 Kennington: tunnel in unit A2 (Gourvenec et al., 2005) Far-field pore pressure measured at x 175 m

Pore pressure: kPa

100

50 x 0 0

5 7 4 1 2 3 6 Horizontal distance from tunnel lining extrados, x : m

Fig. 40. Pore pressures near old tunnels in London Clay (bolted cast iron linings) (measurements made by Tube Lines Ltd) Table 3. Year of construction and tunnel depth for tunnels in Fig. 40 Tunnel Year of construction 1890 1907 1907 1906 1924 Depth of tunnel below ground level: m 15 39 65 28 21

A. B. C. D. E.

Oval Golders Green 1 Golders Green 2 Aldwych Kennington

pressure distribution that would be expected for a tunnel acting as a drain in a uniform deposit of clay. However, case E (at Kennington) shows a markedly different trend. In this case, the pore pressures in the clay even within a few metres from the tunnel are the same as the far-eld value measured 17.5 m from the tunnel, and very close to the tunnel

(0.7 m from the extrados) only a small reduction of pore pressure is seen (Gourvenec et al., 2005). The probable reason for this difference is that in the case of the Kennington tunnel the London Clay at tunnel level is more permeable, resulting in the tunnel lining system being effectively impermeable. The permeability of London Clay can be highly variable, as shown in Fig. 41 (Hight et al., 2007). The data are all from in situ measurements of horizontal permeability, inferred from falling- or rising-head tests in piezometers, from self-boring permeameter tests, or from self-boring pressuremeter tests (Ratnam et al., 2005). It can be seen that at any given depth the variation of permeability can be as much as two orders of magnitude, varying from around 5 3 1011 m/ s to 5 3 109 m/s. This wide variation can in part be explained by the different units of London Clay identied by King (1981). The unit A2, at the bottom of the London Clay stratum, has a higher permeability than the overlying unit A3, as shown by the lines in Fig. 41; moreover, the London

Horizontal permeability: m/s 1 0 B A3 A2 10 London Clay units (King, 1981) 10

12

10

11

10

10

10

10

Near

surfa

ce

Depth below ground level: m

st o

20

wes t

Bottom of London Clay

fb asi n

Clacton Kennington Westminster T5, Heathrow Camden Town Guildford Whitechapel

A3

Warden Point

40

50

Fig. 41. London Clay permeability: in situ tests (Hight et al., 2007)

A2

cen t

30

cen

ral a

Bradwell Wraysbury

and

tral

nd

wes t

Ea

TUNNELLING AND GEOTECHNICS: NEW HORIZONS Clay units to the east of the London basin all exhibit lower permeabilities than in the central and western parts of the basin (Hight et al., 2007). The considerable variation of permeability within the London Clay is the probable explanation of case E (Kennington) showing a markedly different trend in observed long-term pore pressures: see Fig. 40. Only a small reduction of pore pressure was observed very close to the tunnel (0.7 m from the extrados). This is because the tunnel was in the A2 unit, for which the horizontal permeability inferred from self-boring permeameter tests was 2 3 109 m/s; consequently the tunnel lining was relatively impermeable in relation to the surrounding soil (Gourvenec et al., 2005). Field measurements of long-term settlements Field measurements of long-term settlements above tunnels in clays are comparatively rare, not least because on most projects monitoring tends to cease soon after completion of the tunnel. Evidence that the ground surface continues to settle after tunnel excavation in clays has been illustrated by Peck (1969), OReilly et al. (1991), Lake et al. (1992), Shirlaw (1995), Bowers et al. (1996), Nyren (1998) and Harris (2002b). Post-construction settlements were reviewed by Mair & Taylor (1997), who concluded that the major factors inuencing their development are (a) the magnitude and distribution of excess pore pressure generated by the construction of the tunnel (b) the compressibility and permeability of the soil (c) the pore pressure boundary conditions, particularly the permeability of the tunnel lining relative to the permeability of the soil (d ) the initial pore pressure distribution in the ground prior to tunnel construction. OReilly et al. (1991) reported monitoring of longer-term settlements over a period of 11 years for a 3 m diameter tunnel constructed in normally consolidated silty clay in Grimsby; they also measured pore pressures in the ground surrounding the tunnel and found no evidence of reduced pore pressures, even within a few metres of the tunnel. However, in a back-analysis using the nite element method the closest match to the observed consolidation settlements was obtained by assuming the permeability of the combined primary segmental lining and secondary in situ concrete lining to be 5 3 1011 m/s, compared with the permeability of the clay (deduced from in situ constant head tests) of about 1010 m/s (Mair et al., 1992b). It was concluded that the tunnel at Grimsby was acting as a drain, albeit partially, such that the pore pressures in the ground were reduced only very close to the tunnel (probably within a metre or so). It is often observed that tunnels are visibly wet, despite precautions taken in an attempt to make them watertight. In the case of the Jubilee Line Extension tunnels in London, substantial consolidation settlements were observed over tunnels in most locations for periods of up to 5 years following construction: these observations were irrespective of the lining type, whether bolted spheroidal graphite iron or concrete (both of which were grouted), or expanded concrete, or in situ concrete (Harris, 2002b). Measurable consolidation settlements were found at distances up to 100 m from the tunnels. Longer-term settlement monitoring has been undertaken for almost 11 years at two sites in London since completion of the Jubilee Line Extension. The two sites are at St Jamess Park and Elizabeth House, separated by about 1.1 km; their locations are shown on the plan in Fig. 42. At St Jamess Park, as shown in Fig. 43, the westbound (WB)
Th eM all
N River Thames 11 km Treasury Building

715
Elizabeth House

EB WB

St Jamess Park Lake

Houses of Parliament

County Hall

Waterloo Station

Fig. 42. Jubilee Line Extension: locations of long-term settlement monitoring

and eastbound (EB) tunnel axes are located at approximately 31 m and 20.5 m below the ground surface respectively. The 4.85 m OD tunnels are lined with 200 mm thick expanded precast concrete segments: these were not grouted. Detailed descriptions of the St Jamess Park greeneld instrumented site and the tunnel excavations beneath are given by Nyren (1998). Further details of the ground conditions, the tunnelling methodology and the observed ground movements during tunnel construction are given by Standing & Burland (2006) and Dimmock & Mair (2006a, 2006b). As shown in Fig. 43, piezometer measurements at the site indicate initial pore pressures (prior to tunnelling) close to hydrostatic conditions in the London Clay above and around the positions of the tunnels to be constructed: these are consistent with measurements at the nearby Westminster site (Higgins et al., 1996). (Underdrainage below about 35 m and in the underlying Lambeth Group is probable, as observed at Westminster, although piezometers were not installed at depths at St Jamess Park to verify this.) Figure 43 shows post-construction settlement measurements at St Jamess Park (Standing, personal communication, 2006), taken at a depth of 5 m below ground level (to eliminate seasonal effects observed closer to the ground surface). Two sets of data are shown, each directly above each one of the tunnels. The westbound tunnel was constructed rst, followed by the eastbound tunnel about 8 months later. Consolidation settlements only are shown in Fig. 43: that is, settlements occurring during construction of the two tunnels are not shown. The time axis is measured from the completion of the westbound tunnel; the change in gradient of the settlements shown in Fig. 43 can be seen after 8 months, reecting the change occurring after completion of the eastbound tunnel. It is evident that considerable consolidation settlement has occurred since tunnel construction, approaching 80 mm after 11 years, with the rate of settlement remaining almost constant (with some indication of this beginning to reduce after about 10 years). It is also evident that the magnitudes of the consolidation settlements above each tunnel are almost identical. Elizabeth House is a reinforced-concrete-framed ten-storey building, which was closely monitored during construction of the Jubilee Line Extension tunnels beneath it (Standing, 2001). The building is shown in Fig. 44. Prior to tunnelling, Class A predictions (as dened by Lambe, 1973) of the settlements caused by tunnel construction were made: full details of the building construction, the tunnels and the method of prediction are given by Mair & Taylor (2001). The building has two levels of basement, and is founded on a 1.2 m reinforced concrete raft; the base of the raft is in Thames Gravel, a short distance above the interface with the underlying London Clay, as shown in Fig. 45. Mair & Taylor (2001) concluded that the building would respond to tunnel

716
St Jamess Park

MAIR
Initial pore pressure: kPa 0 0 St Jamess Park London Clay 10 EB Westminster data (Higgins et al., 1996) 100 200 300

Depth: m bgl

20

WB

485 m OD expanded concrete segmental linings

30

Hydrostatic

40 Lambeth Group 0 Settlement data 5 m below ground level

Consolidation settlement: mm

20

40 07/02/2006

60

80

4 6 8 Time: years from 01/05/1995

10

12

Fig. 43. St Jamess Park tunnels: cross-section, initial pore pressure prior to tunnelling, and settlement measurements since completion (image courtesy J. Standing)

ments taken in the lower basement level of Elizabeth House (Standing, personal communication, 2006). Three sets of data are shown, one above each of the eastbound and westbound tunnels and one above the crossover tunnel between them: the three sets are very similar. Fig. 46 shows a comparison of the consolidation settlements from St Jamess Park and Elizabeth House plotted against time, from which it can be seen that the magnitude of the Elizabeth House settlements is only about 20% of those observed at St Jamess Park. Also shown in Fig. 46 are the variations of the consolidation settlements after 11 years with distance transverse to the tunnels for the two sites: in each case the settlement prole is very wide and exhibits very little curvature. In summary, the eld evidence from St Jamess Park and Elizabeth House indicates the following.
Fig. 44. Elizabeth House: detailed settlement monitoring during and after construction of Jubilee Line Extension

construction almost perfectly exibly, and the subsequent measurements conrmed this to be the case. As shown in Fig. 45, in contrast with the St Jamess Park site, piezometer measurements close to Elizabeth House indicate initial pore pressures (prior to tunnelling) in the London Clay to be signicantly less than hydrostatic, consistent with underdrainage due to deep-level pumping (Hight et al., 1993); the reduced pore pressures may also be partly a result of the presence of existing tunnels in the Waterloo area. The new tunnels to be constructed, for which the axis level was 22 m below the raft foundation, also differed from those at St Jamess Park: the two 5 m eastbound (EB) and westbound (WB) tunnels were lined with sprayed concrete and an in situ reinforced concrete secondary lining; a crossover tunnel was constructed between the eastbound and westbound tunnels, this being lined with sprayed concrete only (further details are given by Standing, 2001). Figure 45 shows post-construction settlement measure-

(a) Substantial long-term consolidation settlements can occur above tunnels constructed in London Clay. (b) The magnitude and rate of settlement are very different for the two sites. The reasons for the difference between the two sites may be attributable to one or more of the four factors identied earlier. Of these, the most obvious differences are (a) the permeability of the tunnel lining relative to the permeability of the soil (b) the initial pore pressure distribution in the ground prior to tunnel construction. The linings at St Jamess Park were expanded precast concrete segments, which were not grouted; at Elizabeth House, in contrast, the linings were sprayed concrete, which mostly had a secondary in situ reinforced concrete lining. The linings at St Jamess Park were therefore likely to be of signicantly higher permeability than at Elizabeth House, and this could be one of the reasons for the substantially larger consolidation settlements. Comparison of Figs 43 and 45 shows that the initial pore pressure distribution (prior to

TUNNELLING AND GEOTECHNICS: NEW HORIZONS

10-storey building 0 Initial pore pressure: kPa 100 200 300 400 0 Hight et al. (1993)

717

London Clay Sprayed concrete linings in situ concrete (EB and WB)

10

20

Depth: m bgl

Hydrostatic 30

EB Lambeth Group 0

Crossover

WB

40

Consolidation settlements: mm

4 8 12 16 20 0 2

Settlement data lower basement level

50

60 21/02/2006

4 6 8 Time: years from 06/03/1996

10

12

Fig. 45. Elizabeth House: cross-section through tunnels, initial pore pressure prior to tunnelling, and settlement measurements since completion (image courtesy J. Standing)

Consolidation settlement: mm

0 20 40 60 80 0 0 20 40 60 80 30 20 10 0 10 20 30 Distance from eastbound tunnel axis: m 40 50 2 4 6 Time: years 8 10 12 4m Tunnel Elizabeth House (lower basement level) St James's Park (5 m below ground level) St James's Park (5 m below ground level) London Clay 2 Lambeth Group 1 Lambeth Group 2 42 Thanet Sand 48 54 Elizabeth House (lower basement level) 27 m 0 Terrace Gravel 6 12 London Clay 1 18

Pore pressure: kPa 0 200 400

Hydrostatic

Depth: m

24 30 36

Consolidation settlement: mm

Fig. 47. Finite element analysis of long-term tunnel behaviour: a parametric study

Fig. 46. Comparison of consolidation settlements at St Jamess Park and Elizabeth House: settlement measurements with time, and transverse distribution after 11 years (Standing, personal communication, 2006)

tunnelling) was close to hydrostatic at St Jamess Park but signicantly less than hydrostatic at Elizabeth House: this too is a likely reason for the differences in consolidation settlements. Finite element parametric study The importance of the relative permeability of the tunnel lining and soil, and its effect on long-term settlements, has been explored by means of a nite element parametric study. The analyses were undertaken in connection with the Crossrail project by Geotechnical Consulting Group as part of a study with which the author has been associated, using the ICFEP program developed by Professor David Potts. Fig. 47

illustrates the problem that was studied. A single 4 m diameter tunnel at a depth of 27 m in London Clay was analysed, representing a bolted cast iron tunnel constructed for the London Underground many years ago, similar to those listed in Table 3. Tunnel construction was modelled two-dimensionally by using the volume loss control method as outlined by Potts & Zdravkovic (2001) to create a volume loss of 1.5%; the segmental nature of the tunnel lining was accounted for by modelling the joints between individual segments. The constitutive soil model adopted was a non-linear elastic, small-strain, MohrCoulomb formulation; details of the pre-yield model and the parameters assumed are given in Appendix 2. The parametric study investigated, among other things, the inuence of the tunnel lining permeability and the soil permeability. Only the effect of varying these two parameters is reported in this lecture. The permeability of the two London Clay layers was initially assumed to be 109 m/s and isotropic. Only the permeability of the 6 m thick London Clay layer (London Clay 2) in which the tunnel is located was varied in the study. All other soil parameters were kept constant, as detailed in Appendix 2. The initial pore water

718

MAIR meable linings this leads to a reduction in permeability in the soil close to the tunnel in the long term. However, in an analysis of a tunnel in London Clay at a depth of 20 m assuming a log law permeability model proposed by Vaughan (1989), Shin et al. (2002) showed that the longterm reduction in permeability of the soil immediately adjacent to the tunnel was very small. As can be seen from Fig. 48, for Case 1 (impermeable lining) the long-term pore pressures return to their original value, because the tunnel lining is fully impermeable. For Case 2 the fully permeable tunnel lining allows steady-state seepage to develop, with a consequent reduction in pore pressure above and to the side of the tunnel. For Case 3, as would be expected, the tunnel lining of nite permeability results in a long-term pore pressure distribution intermediate between the fully impermeable lining (Case 1) and the fully permeable lining (Case 2). Of particular interest are Cases 4 and 5, for which the tunnel lining is assumed to be fully permeable, but the permeability of the tunnel soil layer is changed. For Case 4 the assumed anisotropy of permeability (kh /kv 4) results in only a small change in the long-term pore pressure prole above the tunnel compared with Case 2, but there is a signicant reduction to the side of the tunnel, extending to a considerable distance. The effect of changing the permeability of the tunnel soil layer is even more marked for Case 5, in which the same anisotropy is assumed (kh /kv 4) but the permeability is increased by a factor of 5. This results in a substantial reduction in long-term pore pressure both above and to the side of the tunnel. The corresponding long-term settlement proles at the ground surface (following tunnel construction) for the ve cases are shown in Fig. 49. For Case 1 a very small heave is predicted, corresponding to the swelling that occurs as the negative excess pore pressures generated by tunnel construction dissipate, with the pore pressures returning to their original values. For Case 2 the reduction of long-term pore pressure associated with the fully permeable lining results in a maximum long-term settlement of 40 mm, with discernible settlement extending to a distance of about 60 m from the tunnel. As would be expected, Case 3 results in a long-term settlement prole intermediate between Cases 1 and 2. Cases 4 and 5 are of considerable signicance. Simply varying the permeability of the tunnel soil layer (either in magnitude or by assuming anisotropy) has a substantial effect on both the magnitude and distribution of the longterm settlement. For Case 5 the maximum long-term settlement increases to almost 80 mm, and the settlement trough is very wide, extending to a distance of 100 m. The ve
50 Pore water pressure: kPa 50 100 150 200 0 0 Initial conditions and Case 1 End of tunnel construction Case 2 6 Case 3 Case 4 Case 5

pressure prole, prior to tunnel construction, was slightly underdrained in the London Clay; the water table in the Thanet Sand was specied to be equivalent to hydrostatic from a depth of 27 m, as shown in Fig. 47. The coefcient of effective horizontal pressure at rest, K0 , was assumed to increase linearly from 0.5 at a depth of 6 m to a maximum of 1.5 at a depth of 12 m, and then decrease linearly to 1.0 at a depth of 24 m; below 24 m K0 was assumed to be constant at 1.0. The pore pressure prole generated immediately following tunnel construction (assumed to be a rapid, undrained process) is shown in Fig. 48. Above the tunnel crown, for a distance of about 7 m, and a similar distance to the side of the tunnel, there is a substantial reduction of pore pressure. (A slight rise in pore pressure is also seen at higher levels above the tunnel crown.) Fig. 48 also shows the long-term pore pressure for the following ve cases, where only the permeability of the tunnel lining or the tunnel soil layer (London Clay 2) is varied. (a) Case 1: impermeable tunnel lining. Tunnel soil layer: k 109 m/s, isotropic. (b) Case 2: fully permeable tunnel lining. Tunnel soil layer: k 109 m/s, isotropic. (c) Case 3: Tunnel lining permeability: 5 3 1011 m/s. Tunnel soil layer: k 109 m/s, isotropic. (d ) Case 4: fully permeable tunnel lining. Tunnel soil layer: kv 109 m/s, anisotropic permeability kh /kv 4. (e) Case 5: fully permeable tunnel lining. Tunnel soil layer: kv 5 3 109 m/s, anisotropic permeability kh /kv 4. An impermeable lining (Case 1) was modelled by prescribing the ow rate into the tunnel as zero throughout the analysis. A fully permeable lining (Cases 2, 4 and 5) was modelled by prescribing the pore water pressure u on the tunnel boundary as zero. Immediately after tunnel excavation, however, suctions exist in the clay adjacent to the soil, and in such cases prescribing u 0 will allow ow of water from the tunnel into the soil. This problem was overcome by setting a special boundary condition that maintained a noow boundary if the pore pressure at any point around the tunnel was less than zero (see, for example, Shin et al., 2002). A lining with nite permeability was modelled by the same method as adopted by Shin et al. (2002), in which a combination of structural beam elements and solid elements was used. For this parametric study constant values of permeability for the London Clay were assumed for the various cases analysed. In reality the permeability reduces with increasing effective stress level (Vaughan, 1989). In the case of per-

Pore water pressure: kPa

Depth: m

250 200 150

12

18 100 50 0 100 24 80 60 40 20

Distance from tunnel centreline: m

Fig. 48. Pore pressure proles on the centreline above the tunnel and at axis level: immediately after tunnel construction, and in the long term for Cases 15

TUNNELLING AND GEOTECHNICS: NEW HORIZONS

Distance from tunnel centreline: m 0 0 20 40 60 80 100

719

Vertical displacement: mm

200

400 Case 1 600 Case 2 Case 3 800 Case 4 Case 5 1000

Fig. 49. Inuence of soil and tunnel lining permeability on long-term settlement proles: Cases 15

cases show that by simply varying the permeabilities of the tunnel lining and the tunnel soil layer within reasonable, credible bounds, with no other parameter being changed, the magnitude of the maximum long-term settlement varies from zero to 80 mm: the width of the settlement trough correspondingly increases. The long-term distortion of the tunnel is also affected by the assumed permeabilities of the tunnel lining and the tunnel soil layer. There is a general tendency for a tunnel lining in clay soils to squat with time, as shown in Fig. 39, with a reduction in the vertical diameter (v ) and a corresponding increase in the horizontal diameter (h ). The changes in vertical and horizontal diameter with time for the ve cases analysed for the parametric study are shown in Fig. 50. For Case 1 there is a very slight (but almost negligible) long-term increase in vertical diameter and a corresponding reduction in horizontal diameter. For Cases 25 the tunnel squats, with the reduction in vertical diameter approximately matching the increase in horizontal diameter. The magnitude of the tunnel squat shown in Fig. 50 follows the same trend as the maximum long-term settlement illustrated in Fig. 49. The squat increases simply as a result of the change in permeability of the tunnel lining or of the tunnel soil layer, and varies from zero to around 20 mm depending on the assumptions made. Field measurements of differences from circularity include construction effects as well as long-term consolidation effects; 20 mm squat (corresponding to v /D 0.5%) is sometimes observed in old tunnels in London Clay, and in some cases can be more; the differences observed in the eld are probably a consequence of different lining or tunnel soil layer permeabilities as well as differences in construction tolerances. Wongsaroj (2005) conducted three-dimensional and two250 200 150 100 50 0 500001 100 150 200 250

dimensional nite element analyses of shield tunnel construction in London Clay, using ABAQUS and a critical-state model; many factors were explored, including stiffness anisotropy, inuence of K0 , initial pore pressure conditions and the inuence of tunnel and soil permeability (Wongsaroj, 2005; Wongsaroj et al., 2006). An extensive parametric study was undertaken of the inuence of tunnel lining and soil permeability on long-term ground movements; similar assumptions were made regarding the tunnel boundary ow conditions as in Shin et al. (2002), referred to earlier. As shown in Fig. 51, Wongsaroj (2005) dened a dimensionless settlement as DS imp perm imp (15)

where is the maximum long-term settlement for a particular case, imp is the maximum long-term settlement for a fully impermeable tunnel lining, and perm is the maximum long-term settlement for a fully permeable tunnel lining. For a fully impermeable tunnel lining, where there is no ow, DS 0, whereas for a fully permeable tunnel lining DS 1. Wongsaroj (2005) also expressed the permeability of the tunnel lining relative to that of the soil in terms of the relative permeability RP, a dimensionless number dened as RP k lining C k soil tL (16)

where klining is the permeability of the tunnel lining; ksoil is : the permeability of the soil (ksoil (kv .kh )0 5 in cases of anisotropic permeability); C is the clay cover above the tunnel crown; and tL is the thickness of the tunnel lining. Figure 52 shows Wongsarojs results plotted in terms of
imp

Diametrical distortion: mm

Case 1 Case 2 Case 3 Case 4 Case 5 001 01 1 10 100

perm

Time since tunnel construction: years DS

Fig. 50. Inuence of soil and tunnel lining permeability on timedependent diametral distortion of tunnel: Cases 15 (increase in horizontal diameter shown positive, decrease in vertical diameter shown negative)

imp perm imp

Fully permeable lining Impermeable lining Finite lining permeability

Fig. 51. Denition of dimensionless settlement DS (Wongsaroj, 2005)

720
00001 0 02 04 0001

MAIR
Relative permeability, RP 001 01 1 10 100 1000 Wongsaroj (2005) This study (tL 02 m)

DS

06 08 10

Impermeable

Fully permeable

Fig. 52. Inuence of tunnel lining permeability on maximum predicted surface settlements: see Fig. 51 and equation (15) for denitions of dimensionless settlement DS, and equation (16) for denition of relative permeability RP)

DS and RP dened above. They all fall within a relatively narrow band, and the results of the parametric study described in this paper also fall within the same band (the equivalent lining thickness tL was 200 mm). From Fig. 52 it can be seen that DS 0, indicating an impermeable system, for RP , 0.1; and DS 1, indicating a fully permeable system, for RP . 100. The results of the ICFEP nite element analyses, described above, shown in Fig. 52 (as the points This study), are in reasonably good agreement with Wongsarojs results. It is clearly a simplication to consider the tunnel lining as uniformly permeable. In reality it is more likely that there are specic leaks, perhaps associated with segmental lining joints (although in most cases, except for expanded linings, the linings are grouted). In the case of in situ concrete linings, shrinkage cracks and construction joints can provide leakage paths. Further research is needed to investigate the inuence of leaks at specic locations in a tunnel lining. Nevertheless, the results summarised in Fig. 52 are of practical value in enabling engineers to evaluate whether or not the liningsoil system is likely to be impermeable or permeable. Summary The pore pressure measurements around old tunnels in London Clay presented earlier indicate that in general the tunnels act as drains, but not in all cases. This is probably the case for tunnels in many types of clay. Many parameters inuence the long-term behaviour, but the key factors are (a) the relative permeability of the tunnel lining and soil, RP: if RP , 0.1 the tunnel lining system is effectively impermeable, whereas if RP . 100 the tunnel lining system is effectively fully permeable (b) the degree of anisotropy of the soil permeability and its variability (c) the initial pore pressure prior to tunnelling. It is clear from the eld evidence from St Jamess Park and Elizabeth House in London, as well as from the parametric study presented in this lecture and from the analyses by Wongsaroj (2005), that long-term settlement associated with tunnels in clays can be appreciable, and can extend to large distances from the tunnel (at least 100 m in some cases). Tunnel linings also squat more when there is larger long-term settlement. Although the magnitude of long-term settlement can be appreciable, the settlement proles are generally very wide, with consequent curvature and differential settlements being generally small: hence the potential damage to buildings and services caused by longterm settlements is likely to be of little consequence in most cases. However, in cases of multiple tunnel construction it

may well be of importance to understand how time-dependent settlements caused by construction of an earlier tunnel might affect the ground through which a later tunnel is to be constructed. A good understanding of the permeability characteristics of the ground (including the degree of anisotropy), together with sophisticated nite element analysis, is essential for realistic prediction of long-term settlement associated with tunnels.

EFFECTS OF TUNNELLING ON BURIED PIPES Analytical solution and proposed new design approach In urban environments underground congestion is becoming increasingly important to tunnel designers. Fig. 53 illustrates the typical variety of underground objects to be considered when contemplating construction of a tunnel: pile foundations (both existing and under construction), other tunnels, and pipelines for services. Estimating the effects of tunnelling on pipelines can be important (see Fig. 54), especially when the infrastructure is old and vulnerable: this generally receives less attention than the assessment of

Fig. 53. Underground congestion in the urban environment (courtesy Keller)

TUNNELLING AND GEOTECHNICS: NEW HORIZONS

Surface settlement GL CL x zp Smax Deformed pipeline z0 i

721

VL Tunnel

VL Smax

Volume loss Maximum settlement Depth to tunnel axis Offset from tunnel centreline (a) Trough width parameter

Fig. 54. Deformation of a pipeline due to tunnelling

i z0 x

tunnelling effects on buildings. Different approaches have been followed, most of which have been based on approximating the problem to Winkler-based elasticity models (e.g. Attewell et al., 1986). Limited validation of such models has been obtained from laboratory experiments and analysis of a number of case histories (ORourke & Trautmann, 1982; Yeates, 1984; Takagi & Nishio, 1984; Bracegirdle et al., 1996). Most of this work, however, relates to small-diameter pipes. A problem with analytical models based on Winkler springs is that designers face a difculty with selection of appropriate values for the subgrade modulus or coefcient of subgrade reaction, especially bearing in mind the dependence of these parameters on pipe diameter. Recent work at Cambridge University has focused on undertaking centrifuge model tests and developing new analytical solutions of pipeline response to tunnelling in terms of continuum elasticity (Vorster, 2005; Klar et al., 2005a; Vorster et al., 2005); monitoring eld performance of large-diameter pipes affected by tunnelling has also been undertaken (Vorster, 2005; Vorster et al., 2006). A closedform solution for a pipe in a continuous elastic medium affected by tunnelling, developed by Klar et al. (2005a), is expressed in terms of the parameters dened in Fig. 55. For a pipe with its axis at a depth zp affected by a tunnel with its axis at a depth z0 , it may for convenience be assumed that the greeneld settlement (ignoring the presence of the pipe) at the level of the pipe is Gaussian. The relevant soil parameters are: VL , the volume loss associated with the tunnelling; Smax , the maximum soil settlement at a level corresponding to the axis of the pipe, and i, the trough width parameter. In addition, elastic soil parameters are Youngs modulus Es and Poissons ratio . For a continuous pipeline the relevant parameters are: the bending stiffness Ep Ip (where Ep is Youngs modulus and Ip is the second moment of area), the axial stiffness Ep A (where A is the cross-sectional area), the outer pipe radius ro, and M, bending moment induced in the pipe. A closed-form solution for the maximum sagging and hogging bending moments induced in a continuous pipeline by tunnelling is shown in Fig. 56: this was derived by Klar et al. (2005a) using an elastic continuum method employing Mindlins solution (Greens function). Normalised bending moment is plotted against relative pipesoil bending stiffness R on a logarithmic scale. The bending moment M is normalised as Mn , dened as Mn Mi2 Ep I p Smax (17)

GL

CL zp

EpIp, EpA, M, ro

VL

EpIp

Bending stiffness

EpA Axial stiffness ro Pipe outer radius zp M Pipe embedment depth Bending moment (b)

Fig. 55. Denitions of parameters for a pipeline affected by tunnelling: (a) subsurface soil settlement in the greeneld; (b) continuous pipeline
12

Normalised bending moment, Mn Mi 2/EpIpSmax

Infinitely flexible behaviour 10 08 06 04 02 0 001 Sagging Overestimation if assume infinite flexibility

Hogging Increasingly stiff 01 1 10 100 Relative pipesoil bending stiffness, R EpIp/Esroi 3

Fig. 56. Pipesoil interaction using continuum elasticity (Vorster et al., 2005)

where Ep Ip is the pipe bending stiffness, and Smax and i are the greeneld settlement trough parameters at the level of

the pipe axis (see Fig. 55). Innitely exible behaviour for settlement described in a Gaussian form corresponds to the maximum normalised sagging moment Mn 1 (and the maximum normalised hogging moment Mn 0.45). The relative pipesoil bending stiffness R is dened as

722

MAIR Ep I p Es ro i3 (18) shear strain a along the pipe between +2.5i and 2.5i; this is also given in Vorster et al. (2005). For pipelines near to the ground surface, it can be shown that the expression leads to a simplied approximation for a , useful for preliminary design purposes, given by a 0:5 VL z0 =D T 2 (19)

where ro is the pipe outer radius and Es is the Youngs modulus of the soil around the pipe. Key assumptions in the derivation of this closed-form solution are as follows. (a) The pipe is buried in homogeneous soil. (b) The pipe is always in contact with the soil (i.e. no separation occurs). (c) The presence of the pipe does not affect the tunnel. (d ) The soil response to loading at pipe level is not affected by the tunnel. (e) The greeneld soil displacement at pipe level is given by the Gaussian equation (see equation (12) and Fig. 55). The closed-form solution shown in Fig. 56 was derived for the greeneld settlement being Gaussian, but the method can be applied to any shape function used to describe the settlement curve (Vorster et al., 2005). It can be seen from Fig. 56 that the bending moments can be signicantly overestimated if the pipe is assumed to be innitely exiblethat is, if it is assumed to follow the greeneld settlement prole, when it in fact reacts in a stiffer manner. The assumption of innitely exible behaviour is valid only for R values less than about 0.1. A pipeline may behave exibly under a given set of circumstances, but stify in another depending on its associated value of R. For smaller values of R (, 0.1) it is reasonable to disregard pipesoil interaction and simply assume that the pipe follows the curvature of the soil; for larger values, this will signicantly overestimate the bending moment, and interaction analysis is required. It is important to note that the scale for R in Fig. 56 is logarithmic. Therefore, if a reasonable estimate can be made for Es, R can be determined with sufcient accuracy for practical purposes. Many pipes are embedded in granular soils, often because they are installed in trenches that have been lled with compacted sand or gravel. As with all soils, there is a degradation of stiffness of such soils with strain level; typical data for Toyoura and Ticino sands are shown in Fig. 57 (Tatsuoka et al., 1997), showing the secant shear modulus Gsec measured in triaxial compression (TC) and plane strain compression (PSC) tests at two different levels of conning stress (comparable to the stress levels applicable to many shallow pipelines). Jovicic & Coop (1997) also provide useful data on the stiffness of coarse-grained soils at small strains. Vorster (2005) derived an expression for the average soil
TC 140 Toyoura sand Ticino sand PSC Conf. pressure 784 kPa 49 kPa

where VL is the tunnel volume loss, and z0 and D T are the depth and diameter of the tunnel respectively. The average soil shear strain increases with volume loss VL , and hence the value of the Youngs modulus Es correspondingly reduces, in the manner shown in Fig. 57. It could be expected, therefore, that the relative pipesoil bending stiffness R would increase with increasing tunnel volume loss; from Fig. 56 it can be seen that the normalised bending moment Mn would correspondingly reduce. A design procedure based on an approach proposed by Vorster et al. (2005) is as follows. (a) Establish the likely greeneld soil displacements at pipe level (volume loss VL, trough width parameter i). (b) Estimate the average soil shear strain at pipe level (using equation (19)) and hence an appropriate soil stiffness Es . (c) Calculate the relative pipesoil bending stiffness R from equation (18). (d ) Calculate the maximum bending moments (and resulting pipe bending strain) using the interaction diagram in Fig. 56. Vorster (2005) showed that the estimation of only bending strain for cases where R . 0.3 provides a conservative estimate of the maximum tensile pipe strain. Where R , 0.3 the combination of axial and bending tensile strains in the hogging location is likely to produce the critical tensile strain for which the pipe should be designed. In the latter case the method should be supplemented by estimates of axial strain (e.g. Attewell et al., 1986; Bracegirdle et al., 1996) to nd the maximum tensile strain for which the pipe should be designed. Validation by centrifuge model tests A series of centrifuge model tests was undertaken by Vorster (2005) on the Cambridge 8 m diameter centrifuge to validate this proposed design procedure, and to explore in detail the mechanisms of pipesoil interaction associated with tunnel-induced ground movements (Vorster et al., 2005b). Fig. 58 shows the layout of the test arrangement. The centrifuge tests were undertaken at 75g in dry sand in which a model tunnel of diameter DT 60 mm was used to represent a 4.5 m diameter tunnel at full scale. Pipes of different diameter Dp and stiffness (Ep Ip and Ep A) were tested at different geometries, varying the cover of the pipe, Cp , and the distance above the tunnel, H. The tunnel comprised a hollow central brass mandrel over which a latex membrane was tted, such that a known volume of water in the annulus between the membrane and the mandrel could be extracted, thereby inducing volume loss. Settlements of the ground surface at distance from the pipe were measured with lasers and LVDTs; settlements of the pipe and of the ground at distance from the pipe (at pipe invert level) were measured using LVDTs with extensometers. Soil stresses and pipe/soil contact pressures were also monitored using miniature stress cells. Fig. 59 shows a centrifuge model during preparation, in which the tunnel and a strain-gauged pipe can be seen prior to placing the sand. Three different model pipes were tested at 1 : 75 scale (one of acrylic, two of

Secant shear modulus, G: MPa

120 100 80 60 40 20
4 3

10

10

10

10

100

Shear strain: %

Fig. 57. Stiffness measurements for Toyoura and Ticino sand (Tatsuoka et al., 1997)

TUNNELLING AND GEOTECHNICS: NEW HORIZONS

50 100 100 100 LVDTs

723

Cp Model pipeline (DP) H z0

Miniextensometers

Model tunnel (DT) 710 LVDT positions

357

Plan view

100 357 100

Test series varied Cp, Dp, H and tunnel volume loss

Fig. 58. Centrifuge model testing of the effect of tunnelling on pipelines (Vorster, 2005)

Pipe

Strong box

Extensometers and strain gauges

Tunnel

35 mm OD aluminium model pipeline

Fig. 59. Centrifuge model preparation (Vorster, 2005)

aluminium alloy), equivalent at full scale to cast iron pipes of 0.7 m and 1.2 m in diameter and a steel pipe of 2.7 m diameter, as shown in Table 4. Figure 60 shows measurements of bending moments for tests on two pipes of different stiffness but with identical

test geometries (pipe diameter Dp 16 mm, Cp /Dp 3, H/DT 0.93, z0 /DT 2.5). Pipe 1 was made of acrylic and pipe 2 of aluminium alloy; pipe 2 had a bending stiffness Ep Ip 16 times that of pipe 1. Normalised bending moment M is plotted against x/i, where x is the horizontal distance

724

MAIR
Table 4. Model pipes used in centrifuge tests at 75g and equivalent prototype pipes in terms of stiffness (Vorster, 2005) Model scale Pipe 1 Pipe 2 Pipe 3 16 mm OD 3 2 mm Acrylic 16 mm OD 3 1.22 mm Aluminium 35 mm OD 3 1.6 mm Aluminium Equivalent prototype pipe 0.73 m OD 3 21.6 mm (28 in OD 3 0.85 in) Cast iron 1.22 m OD 3 34.9 mm (48 in OD 3 1.375 in) Cast iron 2.7 m OD 3 20 mm Steel

M* 06 04 02 Infinitely flexible 04 02 06

M* Infinitely flexible

x/i

0 02 02 04 06 08 10 (a)

x/i

x/i

0 8 6 4 2 02 04 06 08 0 2 4 VL VL VL VL VL (b) 6 04% 1% 2% 4% 6% 8

4 VL VL VL VL VL

6 04% 1% 2% 4% 6%

Mn M*

Mi /EpIpSmax Mn /(Mn)max

10

Fig. 60. Centrifuge modelsinuence of pipe stiffness and tunnel volume loss on bending moments: (a) pipe 1 (acrylic); (b) pipe 2 (aluminium alloy), 16(Ep I p )pipe1

measured from the tunnel centreline, and i is the settlement trough width parameter for the greeneld settlement; M is dened as Mn /Mnmax, with Mn as dened in equation (17) and Mnmax equal to the maximum value of Mn if the pipe reacts innitely exibly, following the soil curvature perfectly. The innitely exible response is shown on both plots for comparison. At low volume loss (VL 0.4%) pipe 1 exhibits almost perfectly exible behaviour whereas pipe 2 shows a signicantly stiffer response (the bending moment plot differing markedly from the innitely exible response). At VL 1% pipe 1 begins to show a slightly stiffer response, and at VL 2% and at higher volume losses a much stiffer response is seen, comparable with those for pipe 2. The reason for the response becoming stiffer with increasing volume loss is twofold: rst, the shear strain is increasing, which leads to a reduction in soil stiffness and a consequent increase in the relative pipesoil bending stiffness R; second, the settlement trough width parameter i tends to reduce with increasing volume loss, at least initially (Vorster et al., 2005). This means that, while pipes of very different stiffness may behave differently at low tunnel volume loss, similar normalised behaviour is exhibited at higher volume losses. The proposed design procedure outlined earlier was tested for pipes 1 and 2 (Vorster, 2005) and found to be reasonably accurate at low volume loss (and smaller soil strains), becoming increasingly conservative with increased volume loss (and hence larger soil strains). Fig. 61 shows a comparison of calculated normalised bending moments, using the proposed design procedure, with measured normalised bending moments in the centrifuge test on pipe 1. It can be seen that for volume losses up to around 1.1% the calculated sagging bending moments are in good agreement with the

Measured normalised bending moment, (Mn)exp

10 09 08 07 06 05 04 03 02 01 0 0
VL 11% VL 20% VL 25% VL 35% VL 03% VL 05% VL VL VL VL 11%

VL

05%

VL

03%

20%

25% 35%

Mn (sagging) Mn (hogging) 10

01 02 03 04 05 06 07 08 09 Calculated normalised bending moment, (Mn)calc

Fig. 61. Centrifuge models: comparison of measured and calculated pipe bending moments for different volume losses using proposed design approach (Vorster, 2005)

measurements, but at higher volume losses the design procedure becomes more conservative (the calculated moments exceeding the measured values by a greater margin). The agreement between calculated and measured hogging bending moments remains good for the full range of volume loss (up to 3.5%). The increasing overestimation of the sagging bending moments at higher volume losses can be explained by the local interaction mechanisms illustrated in Fig. 62. Gap formation (mechanism A) may occur as separation of the soil from the invert of the pipe takes place; correspondingly,

x/i

TUNNELLING AND GEOTECHNICS: NEW HORIZONS

GL D B C A C D

725

al., 2005), is then required to avoid being overly conservative, but accurate estimation of greeneld curvature is of lesser importance. For cases where 0.1 , R , 5, good estimation of greeneld soil curvature and pipesoil interaction analysis are necessary.

Fig. 62. Postulated local pipesoil interaction mechanisms in centrifuge tests (Vorster, 2005): A, gap formation; B, local negative downdrag (soil settling more than pipe); C, local positive downdrag (pipe settling more than soil); D, longitudinal pipesoil interaction

as the soil settles more than the pipeline in this region, local negative downdrag (mechanism B) may occur. At other regions the pipeline may settle more than the soil (mechanism C), and longitudinal interaction (mechanism D) may also become signicant for larger ground movements. Mechanism A was observed in the centrifuge tests undertaken by Vorster (2005), particularly at higher volume losses, and the mechanisms B, C and D were postulated, based on the observations: each of these mechanisms affects the simple continuum elasticity assumptions. Gap formation (mechanism A) is illustrated in Fig. 63, in which the response of a miniature Entran stress cell (with a diaphragm of 4 mm diameter and 0.11 mm thickness) at the invert of a pipe is plotted against tunnel volume loss. It can be seen that the total stress reduces with increasing volume loss, with formation of a gap being indicated at a volume loss of about 1.5%. The inuence of some of the local interaction mechanisms illustrated in Fig. 62 can be introduced into analyses, by means of local plasticity (Klar et al., 2005b), but for design purposes the procedure outlined earlier is often sufcient, especially as expected tunnel volume losses in practice are usually small. As a practical guide to estimating the need for taking account of the effect of pipesoil interaction, Vorster (2005) showed that for R , 0.1 the pipesoil system is likely to behave innitely exibly, with the pipe following the curvature of the greeneld soil. Good estimation of the likely greeneld curvature at pipe level is required to ensure realistic analysis, but no pipesoil interaction analysis is required. For R . 5 the pipe is likely to provide signicant resistance to ground movement; pipesoil interaction analysis, using the procedure outlined earlier (and in Vorster et

150 130

Jointed pipelines and eld measurements The foregoing relates to the response of continuous pipelines to tunnelling. In practice many pipelines are jointed, and it is known that different joints have different properties in relation to joint rotation and axial restraint (e.g. Attewell et al., 1986; Maragakis et al., 2003). Current design practice usually assumes that jointed pipelines respond innitely exibly (i.e. they follow the greeneld ground settlement prole); joint rotation and pullout are usually the only design criteria required (e.g. ORourke & Trautman, 1982; Bracegirdle et al., 1996, Finno et al., 2003). Normalised solutions to evaluate pipeline bending moments and joint rotations are given by Klar et al. (2008), taking account of relative pipe soil bending stiffness and relative pipe-joint stiffness. Vorster (2005) also undertook centrifuge model tests on jointed pipes, using the same experimental procedures as outlined earlier for continuous pipes; the model pipe joints had negligible rotational and axial stiffness compared with individual pipe sections. The inuence of joint location in relation to the tunnel centreline was investigated. It was found that jointed pipelines should not necessarily be regarded as innitely exible: depending on the pipe stiffness, and on the condition and location of the joints, jointed pipelines are able to resist ground movement. There may well be cases where pipe strain should also be a design parameter (as for continuous pipes), along with joint rotation and pullout criteria: full details are given by Vorster (2005). Jointed pipelines may in some circumstances behave as continuous pipelines, especially at low tunnel volume losses. Cambridge University undertook eld measurements of the response of a 942 mm diameter high-pressure water main pipeline in Chingford to construction of a 2.47 m diameter tunnel, as shown in Fig. 64. Part of the pipeline was a jointed prestressed concrete-lined steel cylinder (PSC) and part of it was continuously welded steel. Optical bre was used to measure the longitudinal strain induced at the crown of the PSC portion of the pipeline (the novel optical bre sensing technique used is described later in this lecture). Fig. 65 shows how there was a transition in pipe behaviour from being continuous to fully jointed with increasing tunnel volume loss: this occurred when the tensile limit of the mortar joints was exceeded. The eld monitoring conrmed observations of jointed pipeline behaviour observed in the centrifuge tests. Full details of the eld measurements at Chingford are reported by Vorster (2005) and Vorster et al. (2006).

Pressure: kPa

110 90 70 50 30 10 0 1 4 2 3 Volume loss: % 5 Stress cell

Fig. 63. Gap formation beneath pipe in centrifuge test (Vorster, 2005)

Summary The new closed-form solution for continuous pipes has led to a proposed design approach taking into account the reduction of soil stiffness with increasing shear strain as a result of tunnel volume loss. Centrifuge tests have validated the design approach, and have provided new insights into mechanisms of pipesoil interaction. Flexible pipes may become stiffer with increasing volume loss and associated increasing soil shear strain. Pipe strain is the key design criterion for continuous pipelines. Jointed pipelines, for which joint rotation and pullout criteria are important, may exhibit behaviour similar to continuous pipelines, depending on pipe stiffness and joint details.

726
Concrete-lined steel cylinder

MAIR
Continuously welded steel Made Ground 011 m Terrace Gravel Pipe diameter Pipe axis depth 942 mm 113 m 1067 m London Clay Formation 30139 m 1130 m

Tunnel diameter

247 m

Lambeth Group

139209 m

Fig. 64. Field monitoring of pipeline response to tunnelling at Chingford

Continuous 100 50 C B Jointed A

0 50 100 150 200 250 300 185 E D CB A

1493

405 0 579 122 Offset from tunnel centreline: m (a) Jointed

1036

335 53

100 50

Microstrain (tension positive)

Bending of individual pipe sections E A C B

0 50 100 D 150 200 E DC B A

sensing technique is based on Brillouin optical time domain reectometry (BOTDR) (e.g. Horiguchi et al., 1994). Optical bre sensing in general relies on the interaction between a laser light and the glass material in an optical bre. A major advantage of the system is that the sensing bre is standard single optical bre encased with a 900 m plastic cover, as shown in Fig. 66. This is cheap, being approximately 0.1/m at 2006 prices; a more robust cable is used to connect the sensing bre to the monitoring base. Strains and deformations alter the refractive index and geometry of the optical bre material. These changes perturb the intensity, phase and polarisation of the light-wave propagating along the probing bre. The principle is illustrated in Fig. 67. If a pulse of light is launched through the bre the majority travels through, but a small fraction is scattered back. Different components of light power, each with distinctive peaks at certain wavelengths, are identied, as shown in Fig. 67. In the case of Brillouin scattering the frequency of the backscattered light is shifted by an amount linearly proportional to the strain applied at the scattering location. By resolving the backscattered signal in time and frequency a complete strain prole along the full length of the bre can be obtained. A particular advantage of optical bre technology comes from the low propagation losses that can be obtained with a single-mode optical bre. This means that strain can be measured along the full length (up to 10 km) of a suitably installed optical bre by attaching a BOTDR analyser to one end of the bre, as shown in Fig. 68. The system offers the following features.

Microstrain (tension positive)

Joint

Joint

Joint

Joint

Joint

Joint

Joint

Joint

Joint

Joint

300 185

1493

405 0 579 122 Offset from tunnel centreline: m (b)

1036

335 53

Joint

250

Joint

Fig. 65. Observed jointed pipeline response to tunnelling at Chingford: (a) volume loss 1.2%; (b) volume loss 3.8%
Single fibre

ADVANCES IN FIBRE OPTIC TECHNOLOGY FOR FIELD MONITORING A recent subject of research at Cambridge University is a novel technique that uses distributed optical bre strain sensing (Bennett et al., 2006; Klar et al., 2006; Vorster et al., 2006; Mohamad et al., 2007; Mohamad, 2008). This direct measurement of strain is of considerable potential for many geotechnical and structural applications. The distributed strain

09 mm

Plastic coating Optical fibre

Fig. 66. Types of optical bre cable (Type A single bre used for sensing)

TUNNELLING AND GEOTECHNICS: NEW HORIZONS

Launched light Optical fibre Transmitted light

727

Backscattered light

Anti-Stokes components Rayleigh

Stokes components

Brillouin Raman

Brillouin Raman

Temperature dependent

0 Strain and temperature dependent

Wavelength

Fig. 67. Principle of distributed optical bre sensing

The average strain over 1 m is measured every 200 mm. The range over which the system can work is 510 km. The resolution is 30 (0.003%). The sensors are very low cost, since the optical bre is very cheap (although the analyser itself is expensive). (e) The system is almost real time, typically taking 5 25 min per measurement. ( f ) It is possible to link or switch between bres. Figure 69 shows a comparison of continuous strain measurement using BOTDR optical bre in some concrete test piles with measurements from vibrating wire strain gauges (Bennett et al., 2006). The optical bre was pretensioned to 3000 and then attached to the pile reinforcement using epoxy resin. The agreement between the two measurement systems is good. Research at Cambridge has implemented the BOTDR optical bre system in a number of piles and has shown that very detailed information can be gained by continuous strain measurement, compared with measurement at discrete points down the pile (Klar et al., 2006). The BOTDR technique was applied to the monitoring of strains of the Thameslink Tunnel during construction of the new Thameslink 2000 tunnel beneath it, shown in Fig. 70

(a) (b) (c) (d )

(Mohamad, 2008). The Thameslink Tunnel is an old masonry tunnel of external diameter 8.5 m constructed between 1865 and 1868 using the cut and cover method. In 2005 the new twin Thameslink 2000 Tunnels (TL2K) were constructed as part of the Channel Tunnel Rail Links (CTRL) Section 2 Contract 103 (C103). The new tunnels are of 6 m internal diameter (6.5 m OD), and the northbound tunnel passes underneath the Thameslink Tunnel. with the Midland main line (MML) running at ground level. As shown in Fig. 70, the minimum clearance from the crown of the new tunnel to the invert of the brick-lined tunnel was 3.6 m. Fig. 71 shows the layout of the optical bre, attached at three longitudinal sections (crown and west and east springlines) and ve circumferential sections (CH514 to CH522) spaced over a 60 m length. As shown in Fig. 71, the bre was attached to the brickwork by means of hooks and epoxy resin, having rst been pre-tensioned to 20003000 . Full details of the project are given by Mohamad (2008). Figure 72 shows the general form of the expected strain around the inner face of the old tunnel as a consequence of constructing a new tunnel beneath it: compression around the crown and tension in the walls. A complete record of the development of strain was obtained as the new tunnel approached, was beneath the masonry tunnel, and passed beyond it. The settlement records indicated that the volume loss associated with the new tunnel construction was around 1% and the maximum settlement experienced by the masonry tunnel was 35 mm. Fig. 73 shows an extract from an animation in which the strain variation at ve cross-sections is continuously shown as the new tunnel proceeded southwards beneath the masonry tunnel. The recorded strain is shown at the point when the new tunnel was directly beneath the east wall of the masonry tunnel. Fig. 74 shows the strain when the new tunnel was directly beneath the west wall. In the latter case it can be seen that the maximum tensile strain is 0.25%. However, this was highly localised, and reduced to 0.17% when the new tunnel passed beyond the masonry tunnel. Visual inspection was made at this point, and some visible hairline cracking was indeed observed at the position where the highest tensile strain was recorded. As most cracks in masonry structures tend to appear along the joints of the brickwork, and because the joints had lost some of the mortar, it was difcult to assess whether ne cracks had developed along other sections where high tensile strain was recorded, that is, . 0.1% (Mohamad, 2008).

Scattered light power

Optical pulse input

Detection of Brillouin scattered light

Scattered light power

Stra 2

in 1
BOTDR analyser

Z1 Dis ta

B2

nc

Z2

B1

Lig

fr ht

eq

ue

nc

The frequency shift of the Brillouin scattered light is proportional to the strain

Fig. 68. Application of Brillouin optical time domain reectometry (BOTDR) to distributed strain measurement in optical bre

728

MAIR

Strain: % 0 0 RuFUS project BRE test site Chattenden 001 002 003

Pile group, depth 7 m Minipiles, 143 mm dia. Central pile, 300 mm dia.

Optical fibre

Depth: m
4

Pile 1A - BOTDR Pile 2A - BOTDR VSWG Pile 1B - BOTDR 6 Pile 1A - VWSG Pile 2A - VWSG Pile 1B - VWSG

Fig. 69. Comparison of BOTDR with vibrating-wire strain gauges (VWSG) in piles (Bennett et al., 2006)

In summary, the following can be concluded about BOTDR optical bre sensing. (a) It has shown good comparison with vibrating-wire strain gauge measurement in piles, and has already been used successfully for a number of piling projects. (b) It has provided valuable strain data in the Thameslink masonry tunnel during construction of a new tunnel beneath. (c) The measurement of a continuous strain prole is a big advantage over measurements at discrete locations. (d ) The low cost of installation is attractive. The technique is a promising new development for monitoring of tunnels and many other geotechnical applications. It allows a direct measurement of the tensile strain of masonry material, which can be extremely useful for the process of risk assessment and monitoring of masonry structures inuenced by nearby construction. (d ) CONCLUSIONS A number of new developments in both the theory and practice of tunnelling have been covered. The following principal conclusions can be drawn. (a) Simplied plasticity models are of considerable value for understanding and interpreting the behaviour of deep tunnels in clays, especially in complex ground conditions, where it may be particularly difcult for designers to characterise the ground properties, as has been demonstrated for the tunnels at Bolu. The (b)

(c)

(e)

simplied models have been validated by eld measurements from a range of tunnelling projects. Ground movement control remains critical for tunnelling in urban environments, and earth pressure balance (EPB) tunnelling can routinely achieve low volume losses (,1%) in a wide variety of ground conditions, as has been found for the recently completed CTRL project. Good control of face pressure, through proper control of the excavation chamber pressure, depends on appropriate soil conditioning and screw conveyor operation, about which more is now known. Compensation grouting is an effective method of controlling building response to tunnel construction for most ground conditions. The compensation grouting at Bologna has demonstrated the innovative use of directional drilling to install curved grout tubes. It also demonstrated the successful application of compensation grouting to granular soils, for which there has been generally less experience in comparison to clay soils. Long-term ground movements can be signicant, as demonstrated by the 11 years of measurements taken for the Jubilee Line Extension project in London. Tunnel distortions are related to the same consolidation processes associated with leaking tunnel linings that result in long-term settlements. The magnitude of longterm ground movements and tunnel distortions depends principally on the relative permeability of the tunnel lining and soil, on the degree of anisotropy of the soil permeability (and its spatial variability), and on the initial pore pressure prior to tunnelling. A proposed new design approach for assessing the

TUNNELLING AND GEOTECHNICS: NEW HORIZONS

Reg e Can nts al
Thameslink Tunnel

729

To ventilation shaft

20 00 Tu nn el

N
Junction box

West wall section

Crown section

Fixing points East wall section

Sou thbo und Tha mes link 200 0 Tu nne l

No r th bo un d

Th am es

Regents Canal Bridge

link

Midland main line

Loose CH522 CH520 sections

CH518 83 m (a)

CH516

CH514

Yacht basin box retaining wall CH522

Location of longitudinal crown section Fixture Approximate points location of longitudinal west wall section 3 687 5 4 6 7 8 9 2 Fibre-optic cable 5672 Approximate location of 10 power cables (live) Top of ballast (approximate) 11 823 Approximate location of longitudinal east wall section

CH520 A CH518 A St Pancras Yacht Basin

CH516

CH514 0 30 m 1

7611 mm

(a) Midland main line Waterproof HDP geomembrane Existing clay fill

New fill/formation Retaining wall Yacht basin Steel cup hook Brick lining Tight buffered optical fibre cable TL2K northbound tunnel 65 m OD

(b)

New ground anchors

Existing Thameslink tunnel 3 6 m

Brick surface

Epoxy resin

London Clay (b)

(c)

Fig. 70. New Thameslink 2000 (TL2K) tunnel crossing beneath Victorian brick-lined Thameslink tunnel (Mohamad, 2008): (a) plan; (b) section AA

Fig. 71. Arrangement and xing details of optical bre for strain monitoring of Thameslink tunnel (Mohamad, 2008): (a) general layout; (b) cross-section; (c) xing details

effects of tunnelling on pipes has been presented, taking into account the reduction of soil stiffness with increasing shear strain as a result of tunnel volume loss. Centrifuge tests have validated the design approach and have provided new insights into mechanisms of pipesoil interaction. Flexible pipes may become stiffer with increasing volume loss and associated increasing soil shear strain. Jointed pipelines may exhibit behaviour similar to continuous pipelines, depending on pipe stiffness and joint details. ( f ) BOTDR bre optic technology has been shown to be a highly promising new strain-monitoring technique for tunnelling and many other geotechnical applications. New horizons in tunnelling and geotechnics are ever more challenging: Fig. 75 shows the 15 m diameter Herrenknecht

EPB tunnelling machine that was recently in operation for an urban motorway in Madrid (this is currently the largest EPB tunnelling machine in the world). As tunnels become bigger and more numerous, so the role of geotechnical engineering in such projects will become increasingly important. There are many exciting challenges ahead. ACKNOWLEDGEMENTS There are many individuals and organisations that the author wishes to thank for their help in preparing the lecture and this paper. Most of the material is drawn from research undertaken by the Cambridge Geotechnical Research Group and from consulting projects with Geotechnical Consulting Group (GCG); the author is indebted to both for their constant stimulation and challenges. He is particularly grateful to David Harris, Dr Chris Menkiti, Professor Neil Taylor, Dr

730
West wall Tensile Strain 0 Compression 0 Distance C L New tunnel Crown C L East wall Tensile Compression Old tunnel

MAIR Group (GCG)on the Crossrail project in London, Bologna in Italy, and Bolu in Turkey; I am grateful to my colleagues at GCG and also to Cross London Rail Links Ltd, Italferr of Italy, and KGM and Yuksel-Rendel of Turkey for enabling me to include this material. The assistance of Tube Lines Ltd is also gratefully acknowledged. Finally, the one person who most certainly merits my sincere thanks is my wife Margaret, who has lived for a long time, very patiently, with the thinking and preparation for this lecture.

Fig. 72. Expected strain distribution around inner face of old tunnel as a result of construction of new tunnel beneath

Felix Schroeder, Dr Jamie Standing and Dr Ming Wongsaroj for their invaluable assistance. Dr Peter Bennett, Dr Xavier Borghi, Dr Keith Bowers, Alec Marshall, Dr Andrew Merritt, Hisham Mohamad, Professor Kenichi Soga, Dr Eduard Vorster and Peter Wright were also most helpful. Much of the material in this lecture has drawn on recent research undertaken at Cambridge University, some of which has been sponsored by the Engineering and Physical Sciences Research Council, the Cambridge-MIT Institute and Nishimatsu Construction. The author is grateful to Rail Link Engineering and Union Railways (North) for their permission to include material from eld research undertaken by the Cambridge Geotechnical Research Group on the recently completed CTRL tunnelling project in the London area. This lecture has also drawn on case histories and consulting projects that I have been closely involved with while working with Geotechnical Consulting

APPENDIX 1: SOIL CONDITIONING FOR EPB TUNNELLING General principles

The quantities of soil conditioning agents used in EPB tunnelling machines are expressed in terms of the ratio of the volume of conditioning agent to the volume of ground to be excavated. Polymer injection ratios (PIR) and foam injection ratios (FIR), usually expressed as a percentage, are dened as follows: Vp 3 100 Vs Vf FIR 3 100 Vs PIR (20) (21)

where Vp is the volume of polymer solution, Vf is the volume of foam at atmospheric pressure, and Vs is the volume of soil. The properties of the foam strongly depend on its proportion of air and surfactant solution, which is characterised by the foam expansion ratio (FER), expressed as a percentage:

02

CH522

02

CH520

01 Thameslink Tunnel

01

TL

2K

01 2m 02 CH522

01

West wall

East wall

CH518

: %
CH516

01

CH520

CH518

CH516

01

CH514

02

02

CH514

01

01

0 N 01

01

Fig. 73. Strain recorded when new tunnel is directly beneath east wall of old tunnel at CH520 (Mohamad, 2008); measurements made from west wall at track level around tunnel to east wall at track level (see Fig. 71(b))

TUNNELLING AND GEOTECHNICS: NEW HORIZONS

02 CH522 CH520

731

02

01 Thameslink Tunnel

01

TL 2K

01 2m 02 CH522

01

West wall

East wall

CH518

: %

01

CH520

CH518

CH516

01

CH514

02

CH516

02

CH514

01

01

N 01 01

Fig. 74. Strain recorded when new tunnel is directly beneath west wall of old tunnel at CH516 (Mohamad, 2008); measurements made from west wall at track level around tunnel to east wall at track level (see Fig. 71(b)) Vfl Vp FIR PIR FER Vs

LIR

(23)

In addition the concentration of surfactant and polymer (cs and cp respectively) used in the preparation of the foaming liquid and the polymer solution are dened as cs Vsurf 3 100 Vfl Vpol cp 3 100 Vp (24) (25)

The values of cs and cp affect the properties of the foam and polymer solutions, thereby affecting the properties of the conditioned soil. Limited guidelines on appropriate soil conditioning have been published by EFNARC, based on soil particle distribution only (EFNARC, 2005). Other guidelines have been published by Maidl (1995), Kusakabe et al. (1997), Jancsecz et al. (1999), Milligan (2001) and Merritt (2004). Fig. 75. Herrenknecht EPB machine, 15 m in diameter, for urban motorway tunnel in Madrid Vf FER 3 100 Vfl (22)

Soil conditioning on CTRL Contract 220

Index tests were performed at Cambridge University in advance of tunnelling to assist the contractor in selecting suitable soil conditioning parameters (Mair et al., 2003; Merritt et al., 2003). Full details of the soil conditioning used in practice on CTRL Contract 220 are given by Borghi (2006). The foam injection ratio (FIR) and polymer injection ratio (PIR) for the soil conditioning agents used on Contract 220 are summarised in Fig. 76; further details are given by Borghi (2006) and Borghi & Mair (2006). Average FIR and PIR values are given for all rings (1.50 m excavation) of both the 7.5 km long tunnel drives in each of the

where Vfl is the volume of foaming liquid solution and Vf is the volume of foam. For a given foam agent a range of FER values can be achieved by varying a number of factors in the foam production (Mair et al., 2003; Merritt, 2004). Typically, FER values are around 10 for many conditioning foams. The total liquid injection ratio (LIR) for a foam and/or polymer conditioner injection is given by

732
Thanet Sand (PIR 7%) Foam injection ratio (FIR) Polymer injection ratio (PIR) (FIR 51%)

MAIR
the machine operation with little or no pressure decay during ring build. Observation of the conditioned London Clay at the outlet of the screw conveyor revealed poor mixing when large quantities of foam were used: intermittent discharge of large and stiff clay lumps alternated with gushing of uid and compressed air blows. This heterogeneity is believed to be the result of foam breakdown following sorption of the foaming liquid into the clay, a process also observed in the laboratory (Mair et al., 2003; Merritt et al., 2003).

Lambeth Group (PIR 9%) (FIR 47%)

London Clay (PIR 10%) (FIR 26%) Std dev.

APPENDIX 2: ASSUMPTIONS IN FE ANALYSES FOR PARAMETRIC STUDY OF LONG-TERM SETTLEMENTS Linear elastic parameters
See Table 5.

Non-linear elastic equations and parameters

0 10 20 30 40 50 60 70 80

The tangent shear modulus G and bulk modulus K are given by 3G X c2 1 C1 C2 cos c1 X c2 C2 c1 c2 : sin c1 X c2 p9 2 303 K Y c4 1 C4 C5 cos c3 Y c4 C5 c3 c4 : sin c3 Y c4 p9 2 303 where X log10 and Y log10    (26) (27)

Injection rate: %

Fig. 76. Injection rates for soil-conditioning agents used in EPB tunnelling on CTRL Contract 220 (Borghi & Mair, 2006)

main types of ground conditions. In the Thanet Sand, the average FIR and PIR values were 51% and 7% respectively. The FIR falls within the range 4060 recommended by EFNARC (2005) for sandy soils. In the Lambeth Group, an average FIR of 47% and an average PIR of 9% were measured. The FIR showed a relatively large standard deviation, illustrating the difculty of determining appropriate conditioning treatments in this heterogeneous soil stratum. In many instances FIRs in excess of 200% were used in the Lambeth Group, but no direct benets of such large quantities of foams could be observed (Borghi, 2006). The average FIR used in the London Clay was 26%, that is, around 50% of the values used in the Thanet Sand and in the Lambeth Group. The average PIR was 13% for the rst tunnel, but was reduced to 6% in the second, with an overall average of 10%. The FIR used in the London Clay fell below the range of 3080% recommended by EFNARC for clays. However, analysis of machine data suggested that the principal effect of the foam in clay was mostly that of its liquid phase, and that the mechanisms that make foam a suitable additive in sand cannot be expected to be effective in clayey material (Borghi, 2006). Much lower quantities of foam were used in the London Clay than in the Thanet Sand and the Lambeth Group. PIRs of about 15% or less with little or no foam were found adequate to remould the clay mixtures and allow accurate control of Table 5. Linear elastic parameters assumed Youngs modulus, E Terrace Gravel Thanet Sand Tunnel lining 20.0 MPa 500.0 MPa 100.0 3 106 kPa

Ed 1:732C3 

C6

and the other parameters used in these equations are given in Table 6.

Mohr-Coulomb yield surface parameters

See Table 7.

NOTATION
A A0 a C Cp total face area; cross-sectional area of pipe total openings surface area tunnel radius clay cover above tunnel crown pipe cover

Poissons ratio, 0.2 0.2 0.3

Area, A: m2 33.66 3 103

Second moment of area, I: m4 3.9687 3 105

Table 6. Parameters assumed in equations (26) and (27) C1 London Clay 1 London Clay 2 Lambeth Group 1 Lambeth Group 2 1400.0 1400.0 1400.0 1400.0 C4 London Clay 1 London Clay 2 Lambeth Group 1 Lambeth Group 2 686.0 686.0 686.0 686.0 C2 1270.0 1270.0 1270.0 1270.0 C5 633.0 633.0 633.0 633.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 C3 : % 3 3 3 3 104 104 104 104 c1 1.335 1.335 1.335 1.335 c3 2.069 2.069 2.069 2.069 c2 0.617 0.617 0.617 0.617 c4 0.420 0.420 0.420 0.420 Ed(min) : % 8.66 8.66 8.66 8.66 3 3 3 3 104 104 104 104 Ed(max) : % 0.693 0.693 0.693 0.693 v(max) : % 0.15 0.15 0.15 0.15 Gmin : kPa 2667.0 2667.0 2667.0 2667.0 Kmin : kPa 5000.0 5000.0 5000.0 5000.0

C6 : % 3 3 3 3 103 103 103 103

v(min) : % 5.0 5.0 5.0 5.0 3 3 3 3 103 103 103 103

TUNNELLING AND GEOTECHNICS: NEW HORIZONS

Table 7. MohrCoulomb yield surface and plastic potential parameters Cohesion, c9: kPa 0.0 5.0 5.0 5.0 5.0 0.0 Angle of shearing resistance, 9: degrees 35.0 25.0 25.0 25.0 25.0 40.0 Angle of dilation, 9: degrees 17.5 12.5 12.5 12.5 12.5 0.0

733

Terrace Gravel London Clay 1 London Clay 2 Lambeth Group 1 Lambeth Group 2 Thanet Sand

During pore water pressure dissipation in the long term an angle of dilation, 0.08 has been assumed. c9 cp cs D Dp DT DS El Ep Es Eu fcu G Gsec G0 H Ip i K K0 k kh klining ksoil kv M Mn , M N N P p p9 p10% R RP r ro S Smax su t, tL u Vf Vfl VL Vp VS Vs x, y zp z0 h v imp perm effective cohesion concentration of polymer concentration of surfactant tunnel outside diameter; diameter of tunnel lining pipe diameter tunnel diameter dimensionless settlement Youngs modulus of tunnel lining Youngs modulus of pipe Youngs modulus of soil undrained Youngs modulus shotcrete design cube strength tangent shear modulus secant shear modulus maximum shear modulus distance above tunnel second moment of area of pipe trough width parameter, settlement trough width parameter equivalent spring stiffness; trough width; bulk modulus coefcient of effective horizontal pressure at rest permeability horizontal permeability permeability of tunnel lining permeability of soil vertical permeability bending moment induced in pipe normalised bending moment stability ratio (0 L )/su 0 /su distance of lining behind tunnel face average chamber pressure mean normal effective stress tenth percentile of distribution of chamber pressure p relative pipesoil bending stiffness dimensionless relative permeability radius pipe outer radius settlement maximum settlement undrained shear strength thickness of tunnel lining pore water pressure volume of foam at atmospheric pressure volume of foaming liquid solution volume loss associated with tunnelling; tunnel volume loss volume of polymer solution volume of settlement trough per metre length of tunnel volume of soil horizontal distance measured from tunnel centreline pipe axis depth; pipe embedment depth tunnel axis depth increase in horizontal diameter reduction in vertical diameter radial ground movement/soil deformation; maximum long-term settlement maximum long-term settlement for fully impermeable tunnel lining maximum long-term settlement for fully permeable tunnel lining r 1 a L Li n 9 r v0 v 9 0 9 9 radial ground movement at radius r radial ground movement at tunnel face average soil shear strain opening ratio Poissons ratio pressure on tunnel lining maximum pressure on tunnel lining normal effective stress total radial stress acting at external radius of tunnel lining total overburden pressure at tunnel axis level initial vertical effective stress total overburden pressure at tunnel axis shear stress effective stress friction angle; angle of shearing resistance angle of dilation

REFERENCES
Attewell, P. B., Yeates, J. & Selby, A. R. (1986). Soil movements induced by tunnelling and their effects on pipelines and structures. Glasgow: Blackie. Bennett, P. J., Klar, A., Vorster, T. E. B., Choy, C. K., Mohamed, H., Soga, K., Mair, R. J., Tester, P. & Fernie, R. (2006). Distributed optical bre strain sensing in piles. Proceedings of the international conference on reuse of foundations for urban sites, Watford, pp. 7178. Boone, S. J., Artigiani, E., Shirlaw, J. N., Ginanneschi, R., Leinala, T. & Kochmanova, N. (2005). Use of ground conditioning agents for earth pressure balance machine tunnelling. Proceedings of the AFTES International Congress, Cambery, pp. 313 320. Borghi, F. X. (2006). Lubrication and soil conditioning in pipejacking and tunnelling. PhD thesis, Cambridge University. Borghi, F. X. & Mair, R. J. (2006) Soil conditioning for EPB tunnelling machines in London ground conditions. Tunnels and Tunnelling International, September, 1820. Bowers, K. H., Miller, D. M. & New, B. M. (1996) Ground movement over three years at the Heathrow Express Trial Tunnel. In Geotechnical aspects of underground construction in soft ground (eds R. J. Mair and R. N. Taylor), pp. 647652. Rotterdam: Balkema. Bracegirdle, A., Mair, R. J., Nyren, R. J., & Taylor, R. N. (1996). Criteria for the estimation of damage to services caused by ground movements arising from tunnelling. In Geotechnical aspects of underground construction in soft ground (eds R. J. Mair & R. N. Taylor), pp. 653658. Rotterdam: Balkema. British Tunnelling Society (2005). Closed face tunnelling machines and ground stability: A guideline for best practice. London: Thomas Telford. Clayton, C. R. I., van der Berg, J. P., Heymann, G., Bica, A. V D. . & Hope, V. S. (2002). The performance of pressure cells for sprayed concrete tunnel linings. Geotechnique 52, No. 2, 107 116. Dimmock, P. S. (2003) Tunnelling-induced ground and building movement on the Jubilee Line Extension. PhD thesis, Cambridge University. Dimmock, P. & Mair, R. J. (2006a). Volume loss experienced on two open-face London Clay tunnels. Proc. Instn Civ. Engrs Geotech. Engng 160, No. 1, 311. Dimmock, P. & Mair, R. J. (2006b). Estimating volume loss for

734
open-face tunnels in London Clay. Proc. Instn Civ. Engrs Geotech. Engng 160, No. 1, 1322. EFNARC (2005). Specication and guidelines for the use of specialist products for soft ground tunnelling. Farnham: European Federation for Specialist Construction Chemicals and Concrete Systems, Surrey, UK ([Link]/pdf/[Link]). Finno, R. J., Molnar, K. M. and Rossow, E. C. (2003). Analysis of effects of deep braced excavations on adjacent buried utilities. Report for the US Department of Transportation, No. A450, A464. Northwestern University, Illinois. Gaerber, R. (2003). Design of deep galleries in low permeable saturated porous media. Doctoral thesis, Ecole Polytechnique Federale de Lausanne. Gourvenec, S. M., Mair, R. J., Bolton, M. D. & Soga, K. (2005). Ground conditions around an old tunnel in London Clay. Proc. Instn Civ. Engrs Geotech. Engng 158, No. 1, 2533. Harris, D. (2001). Protective measures. In Building response to tunnelling: Case studies from construction of the Jubilee Line Extension, London (eds J. B. Burland, J. R. Standing and F. M. Jardine), Vol. 1, Projects and methods, pp. 135176. London: Thomas Telford, CIRIA special publicaion 200. Harris, D. (2002a). The Big Ben Clock Tower and the Palace of Westminster. In Building response to tunnelling: Case studies from construction of the Jubilee Line Extension, London (eds J. B. Burland, J. R. Standing and F. M. Jardine), Vol. 2, Case studies, pp. 453508. London: Thomas Telford Harris, D. I. (2002b) Long term settlement following tunnelling in overconsolidated London Clay. In Geotechnical aspects of underground construction in soft ground (eds Kastner, Emeriault, Dias and Guilloux), pp. 393398. Lyon: Specique. Harris, D. I., Mair, R. J., Love, J. P., Taylor, R. N. & Henderson, T. O. (1994). Observations of ground and structure movements for compensation grouting during tunnel construction at Water loo Station. Geotechnique 44, No. 4, 691713. Harris, D. I., Menkiti, C. O., Pooley, A. J. & Stephenson, J. A. (1996). Construction of low-level tunnels below Waterloo Station with compensation grouting for the Jubilee Line Extension. In Technical aspects of underground construction in soft ground (eds R. J. Mair and R. N. Taylor), pp. 361366. Rotterdam: Balkema. Harris, D. I., Mair, R. J., Burland, J. B. & Standing, J. (1999) Compensation grouting to control tilt of Big Ben Clock Tower. In Geotechnical aspects of underground construction in soft ground (eds O. Kusakabe, K. Fujita and Y. Miyazaki), pp. 225 232. Rotterdam: Balkema. Higgins, K. G., Potts, D. M. & Mair, R. J. (1996). Numerical modelling of the inuence of a deep excavation and bored tunnels on the Palace of Westminster clock tower. In Geotechnical aspects of underground construction in soft ground (eds R. J. Mair and R. N. Taylor), pp. 525530. Rotterdam: Balkema. Hight, D. W., Higgins, K. G., Jardine, R. J., Potts, D. M., Pickles, A. R., De Moor, E. K. & Niyrenda, Z. M. (1993). Predicted and measured tunnel distortions associated with construction of Waterloo International Terminal. In Predictive soil mechanics (eds G. T. Houlsby and A. N. Schoeld), pp. 317338. London: Thomas Telford. Hight, D. W., Gasparre, A., Nishimura, S., Minh, N. A., Jardine, R. J. & Coop, M. R. (2007). Characteristics of the London Clay from the Terminal 5 site at Heathrow Airport. Geotechnique 57, No. 1, 318 Hoek, E. & Brown, E. T. (1980) Underground excavations in rock. London: Institute of Mining and Metallurgy. Horiguchi, T., Kurashima, T. & Koyamada, Y. (1994). 1-m Spatial resolution measurement of distributed Brillouin frequency shift in single-mode bers. Proceedings of the NIST/IEEE symposium on optical ber measurements, Boulder, CO, pp. 7376. Jancsecz, S., Krause, R. & Langmaack, L. (1999). Advantages of soil conditioning in shield tunnelling: experiences of LRTS Izmir. Proceedings of the ITA Conference, Oslo, pp. 865875. Jovicic, V. & Coop, M. R. (1997). Stiffness of coarse grained soils at small strains. Geotechnique 47, No. 3, 545561. King, C. (1981). The stratigraphy of the London Basin and associated deposits. Tertiary Research Special Paper 6. Rotterdam: Backhuys.

MAIR
Klar, A., Vorster, T. E. B., Soga, K. & Mair, R. J. (2005a). Soil pipetunnel interaction: comparison between Winkler and elastic continuum solutions. Geotechnique 55, No. 6, 461466. Klar, A., Vorster, T. E. B., Soga, K. & Mair, R. J. (2005b). Continuum solution of soilpipe interaction including local failure. Proc. 11th Int. Conf. of IACMAG: Prediction, Analysis and Design in Geomechanical Applications, Turin, 2, 687694. Klar, A., Bennett, P. J., Soga, K., Mair, R. J., Tester, P., Fernie, R., St John, H. D. & Torp-Peterson, G. (2006). Distributed strain measurement for pile foundations. Proc. Instn Civ. Engrs Geotech. Engng 159, No. 3, 135144. Klar, A., Marshall, A. M., Soga, K. & Mair, R. J. (2008) Tunnelling effects on jointed pipelines. Can. Geotech. J. 45, No. 1, 131 139. Kummerer, C., Thurner, R., Rigazio, A. & Zamagni, A. (2007). Compensation grouting for limiting settlements of two railway bridges induced by a twin-tunnel excavation. Proc. 14th Eur. Conf. Soil Mech. Geotech. Engng, Madrid. Kusakabe, O., Nomoto, T. & Imamura, S. (1997). Geotechnical criteria for selecting mechanised tunnel systems and DMM for tunnelling: Panel discussion. Proc. 14th [Link]. Soil Mech. Geotech. Engng, Hamburg 4, 24392440. Lake, L. M., Rankin, W. J. & Hawley, J. (1992). Prediction and effects of ground movements caused by tunnelling in soft ground beneath urban areas, CIRIA Project Report 30. London: Construction Industry Research and Information Association. Lambe, T. W. (1973). Predictions in soil engineering. Geotechnique 23, No. 2, 149202. Leinala, T., Grabinsky, M., Delmar, R. & Collins, J. R. (2000). Effects of foam soil conditioning on EPBM performance. Proc. North American Tunnelling 00, 514524. Macklin S. R. (1999). The prediction of volume loss due to tunnelling in overconsolidated clay based on geometry and stability number. Ground Engng 32, No. 4, 3033. Maidl, U. (1995). Erweiterung der Einsatzbereiche der Erddruckschilde durch Bodenkonditionierung mit Schaum. PhD thesis, Ruhr University, Bochum (in German). Mair, R. J. (1993). Developments in geotechnical engineering research: application to tunnels and deep excavations. Unwin Memorial Lecture 1992. Proc. Instn Civ. Engrs Civ. Engng 93, No. 1, 2741. Mair, R. J. (1996). General report on settlement effects of bored tunnels. In Geotechnical aspects of underground construction in soft ground (eds R. J. Mair and R. N. Taylor), pp. 4353. Rotterdam: Balkema. Mair, R. J. & Hight, D. W. (1994). Compensation grouting. World Tunnelling, November, 361367. Mair, R. J. & Taylor, R. N. (1993). Predictions of clay behaviour around tunnels using plasticity solutions. In Predictive soil mechanics (eds G. T. Houlsby and A. N. Schoeld), pp. 449 463. London: Thomas Telford. Mair, R. J. & Taylor R. N. (1997). Bored tunnelling in the urban environment: State-of-the-art report and theme lecture. Proc. 14th Int. Conf. Soil Mech. Found. Engng, Hamburg 4, 2353 2385. Mair, R. J. & Taylor, R. N. (2001). Elizabeth House: settlement predictions. Building response to tunnelling. In Case studies from construction of the Jubilee Line Extension, London. Vol. 1: Projects and methods (eds J. B. Burland, J. R. Standing and F. M. Jardine), pp. 195215. London: Thomas Telford, CIRIA special publication 200. Mair, R. J., Gunn, M. J., & OReilly, M. P. (1981). Ground movements around shallow tunnels in soft clay. Proc. 10th Int. Conf. Soil Mech. Found. Engng, Stockholm 1, 323328. Mair, R. J., Taylor, R. N. & Clarke, B. G. (1992a). Repository tunnel construction in deep clay formations. Commission of the European Communities Report EUR 13964 EN. Luxembourg: Ofce for Ofcial Publications of the European Communities. Mair, R. J., Hight, D. W. & Potts, D. M., (1992b). Finite element analyses of settlements above a tunnel in soft ground, TRRL Contractor Report 265. Crowthorne: Transport and Road Research Laboratory. Mair, R. J., Taylor, R. N. & Bracegirdle, A. (1993). Sub-surface settlement proles above tunnels in clays. Geotechnique 43, No. 2, 315320. Mair, R. J., Harris, D. I., Love, J. P., Blakey, D. & Kettle, C.

TUNNELLING AND GEOTECHNICS: NEW HORIZONS

(1994). Compensation grouting to limit settlements during tunnelling at Waterloo Station. Proceedings of Conference Tunnelling 94, London, 279300. Mair, R. J., Taylor, R. N. & Burland, J. B. (1996). Prediction of ground movements and assessment of risk of building damage due to bored tunnelling. In Geotechnical aspects of underground construction in soft ground (eds R. J. Mair and R. N. Taylor), pp. 713718. Rotterdam: Balkema. Mair, R. J., Merritt, A. S., Borghi, F. X., Yamazaki, H. & Minami, T. (2003) Soil conditioning for clay soils. Tunnels and Tunnelling Int., 35, No. 4, 2932. Maragakis, E., Siddhartan, R. & Meis, R. (2003). Static axial behaviour of pipe joints. University of Nevada, Reno, USA ([Link] Menkiti, C. O., Mair, R. J. & Miles, R. (2001a). Tunnelling in complex ground conditions in Bolu, Turkey. Proc. Underground Construction 2001, London, 546558. Menkiti, C. O., Sanders, P., Barr, J., Mair, R. J., Cilingir, M. & James, S. (2001b). Effects of the 12th November 1999 Duzce Earthquake on Stretch 2 of the Gumusova-Gerede Motorway in Turkey. Proc. Int. Road Federation 14th World Road Congress, Paris, CD-ROM. Merritt, A. S. (2004). Soil conditioning for earth pressure balance machines. PhD thesis, Cambridge University. Merritt, A. S. & Mair, R. J. (2006). Mechanics of tunnelling machine screw conveyors: model tests. Geotechnique 56, No. 9, 605615. Merritt, A. S. & Mair, R. J. (2008). Mechanics of tunnelling machine screw conveyors: a theoretical model. Geotechnique 58, No. 2, 7994. Merritt, A. S., Borghi, F. X. & Mair, R. J. (2003). Conditioning of clay soils for earth pressure balance tunnelling machines. Proc. Underground Construction 2003, London, 455466. Milligan, G. W. E. (2000). Lubrication and soil conditioning in tunnelling, pipe jacking and microtunnelling: A state-of-the-art review. London: Geotechnical Consulting Group ([Link]/research/pipejack/[Link]). Milligan, G. W. E. (2001). Soil conditioning and lubricating agents in tunnelling and pipe jacking. Proc. Underground Construction Symp. 2001, London, 105116. Mohamad, H. (2008). Distributed bre optic strain sensing of geotechnical structures. PhD thesis, University of Cambridge. Mohamad, H., Bennett, P. J., Soga, K., Mair, R. J., Lim, C.-S., Knight-Hassell, C. K. & Ow, C. N. (2007). Monitoring tunnel deformation induced by close-proximity bored tunnelling using distributed optical ber strain measurements. Proc. 7th Int. Symp. on Field Measurements in Geomechanics, Boston. Neerdael, B. & De Bruyn, D. (1989). Geotechnical research in the test drift of the HADES underground research facility at Mol. Proceedings of the CED Technical Session on Geomechanics of clays for radioactive waste disposal, Commission of the European Communities Report EUR 12027 EN/FR, pp. 8394. Nyren, R. (1998). Field measurements above twin tunnels in London Clay. PhD thesis, Imperial College, University of London. OCarroll, J. B. (2005). A guide to planning, constructing and supervising earth pressure balance TBM tunnelling. New York: Parsons Brinkerhoff. OReilly, M. P., Mair, R. J. & Alderman, G. H. (1991). Long-term settlements over tunnels: an eleven-year study at Grimsby. Proc. Conf. Tunnelling 91, London, 5564. ORourke, T. D. & Trautmann, C. H. (1982). Buried pipeline response to tunnel ground movements. Proc. Europipe 82 Conf., Basel, 915. ORourke, T. D., Goh, S. H., Menkiti, C. O. & Mair, R. J. (2001). Highway tunnel performance during the 1999 Duzce earthquake. Proc. 15th Int. Conf. Soil Mech. Geotech. Engng, Istanbul 2, 3651368. Palmer, J. H. L. & Belshaw, D. J. (1980). Deformations and pore pressure in the vicinity of a precast, segmented, concrete-lined tunnel in clay. Can. Geotech. J. 17, No. 2, 174184. Panet, M. & Guenot, A. (1982). Analysis of convergence behind the face of a tunnel. Proc. Tunnelling 82, London, 197204. Peck, R. B. (1969). Deep excavations and tunnelling in soft ground. Proc. 7th Int. Conf. Soil Mech. Found. Engng, Mexico City, 225290.

735

Potts, D. M. & Zdravkovic, L. (2001). Finite element analysis in geotechnical engineering: Application. London: Thomas Telford. Poulos, H., Day, P., Valenzuela, L., Crawford, S., Mayne, P., Bolton, M., Tatsuoka, T. & Koseki, J. (2005). Practitioner/academic forum. Proc. 16th Int. Conf. Soil Mech. Geotech. Engng, Osaka 5, 29172936. Rankin, W. J. (1988). Ground movements resulting from urban tunnelling: predictions and effects. In Engineering geology of underground movement, Engineering Geology Special Publication No. 5, 7992. London: Geological Society. Ratnam, S., Soga, K. & Whittle, R. W. (2005). A eld permeability measurement technique using a conventional self-boring pres suremeter. Geotechnique 55, No. 7, 527537. Schmidt, B. (1969). Settlements and ground movements associated with tunnelling in soil. PhD thesis, University of Illinois. Schubert, P., Moggiolli, M., Brandl, H. & Golser, J. (1997) Extraordinary difculties driving the motorway tunnels through Bolu Mountains, Turkey. Felsbau 15, No. 5. Selemetas, D. (2005). The response of full-scale piles and piled structures to tunnelling. PhD thesis, Cambridge University. Shin, J. H., Addenbrooke, T. I. & Potts, D. M. (2002). A numerical study of the effect of groundwater movement on long-term tunnel behaviour. Geotechnique 52, No. 6, 391403. Shirlaw, J. N. (1995). Observed and calculated pore pressures and deformation induced by earth pressure balanced shield. Can. Geotech. J. 32, 181189. Shirlaw, J. N., Ong, J. C. W., Rosser, H. B., Tan, C. G., Osborne, N. H. & Heslop, P. E. (2003). Local settlements and sinkholes due to EPB tunnelling. Proc. Instn Civ. Engrs Geotech. Engng 156, No. 4, 193211. Standing, J. R. (2001). Elizabeth House, Waterloo. In Building response to tunnelling: Case studies from the Jubilee Line Extension, London. Vol. 2, Case studies (eds J. B. Burland, J. R. Standing and F. M. Jardine), pp. 547612. London: Thomas Telford, CIRIA special publication 200. Standing, J. R. & Burland, J. B. (2006). Unexpected tunnelling volume losses in the Westminster area, London. Geotechnique 56, No. 1, 1126. Takagi, N., Shimamura, K. & Nishio, N. (1984). Buried pipe response to adjacent ground movements associated with tunnelling and excavations. Proc. 3rd Int. Conf. on Ground Movements and Structures, Cardiff, 97112. Tatsuoka, F., Jardine, R. J., Lo Presti, D., Di Benedetto, H. & Kodaka, T. (1997). Characterising the pre-failure deformation properties of geomaterials. Proc. 14th Int. Conf. Soil. Mech. Found. Engng, Hamburg, 4, 21292164. Vaughan, P. R. (1989). Non-linearity in seepage problems: theory and eld observation. In De Mello volume: A tribute to Prof. Dr Victor F. B. de Mello, pp. 501515. Sao Paulo, Brasil : Editura Edgard Blucher. Viggiani, G. (2001). Grout intensities. In Building response to tunnelling: Case studies from the Jubilee Line Extension, London Vol. 1: Projects and Methods (eds J. B. Burland, J. R. Standing and F. M. Jardine), pp. 311313. London: Thomas Telford, CIRIA Special Publication 200. Vorster, T. E. B. (2005) The effects of tunnelling on buried pipes. PhD thesis, Cambridge University. Vorster, T. E. B., Klar, A., Soga, K., & Mair, R. J. (2005). Estimating the effects of tunneling on existing pipelines. ASCE J. Geotech. Geoenviron. Engng 131, No. 11, 13991410. Vorster, T. E. B., Soga, K., Mair, R. J., Bennet, P. J., Klar, A. & Choy, C. K. (2006). The use of bre optic sensors to monitor pipeline response to tunnelling. Proc. Geo-Congress 2006, Atlanta, CD-ROM. Ward, W. H. (1969). Yielding of ground and the structural behaviour of linings of different exibility in a tunnel in London Clay. Proc. 7th Int. Conf. Soil Mech. Found. Engng, Mexico City 3, 320325. Ward, W. H. (1978). Ground support for tunnels in weak rocks. Geotechnique 28, No. 2, 133170. Ward, W. H. & Pender, M. J. (1981). Tunnelling in soft ground: General report. Proc. 10th Int. Conf. Soil Mech. Found. Engng, Stockholm 4, 261275. Ward, W. H. & Thomas, H. S. H. (1965). The development of earth loading and deformation in tunnel linings in London Clay. Proc. 6th Int. Conf. Soil Mech. Found. Engng, Toronto 2, 432436.

736
Wong, H., Trompille, V., Subrin, D. & Guilloux, A. (1999). Tunnel face reinforced by longitudinal bolts: analytical model and in situ data. In Geotechnical aspects of underground construction in soft ground (eds O. Kusakabe, K. Fujita and Y. Miyazaki), pp. 457462. Rotterdam: Balkema. Wongsaroj, J. (2005). Three-dimensional nite element analysis of short- and long-term ground response to open face tunnelling in stiff clay. PhD thesis, Cambridge University. Wongsaroj, J., Borghi, F. X., Soga, K., Mair, R. J., Sugiyama, T., Hagiwara, T. & Bowers, K. J. (2005). Effect of TBM driving parameters on ground surface movements: Channel Tunnel Rail Link Contract 220. In Geotechnical aspects of underground construction in soft ground (eds Bakker et al.), pp. 335341. London: Taylor & Francis Group. Wongsaroj, J., Soga, K. & Mair, R. J. (2006). Modelling of longterm ground response to tunnelling under St James Park London. Geotechnique 57, No. 1, 7590. Woods, E., Battye, G., Bowers, K. & Mimnagh, F. (2007). Channel Tunnel Rail Link Section 2: London Tunnels. Proc. Inst. Civ. Engrs Civ. Engng 160, special issue 2, 2428. Yeates, J. (1984). The response of buried pipelines to ground movements caused by tunnelling in soil. Proc. 3rd Int. Conf. Ground Movements and Structures, Cardiff, 129144.

MAIR I rst met him while we were both research students toiling with our PhD work. This ability has allowed him to remain at the forefront of the tunnelling eld throughout his career. I presume that here in the audience there are a substantial number of people who are or have been involved in tunnelling, and I am sure that we would all agree that tunnelling is a rather messy business. However, I think it would be difcult to infer this from this Rankine lecture, in which the important issues have been shrewdly identied, elegantly analysed, and usefully concluded. Some time ago, I came across a collection of short essays discussing the work of eminent Cambridge scientists from William Gilbert in the sixteenth century to our times. It was interesting to detect a common thread in the tradition: an ability to reduce complex phenomena, through illuminating insights, to rational theories and frameworks. This Rankine lecture is a clear proof that this tradition is very much alive today. We have seen, for instance, how apparently simple models can be usefully applied to the understanding and solution of stability problems when tunnelling in extremely complex geological conditions. This is refreshing at a time when three-dimensional analyses often plucked from thin air seem to be replacing judgement. The combination of theoretical insight, eld observations, laboratory testing and sound engineering demonstrates without any doubt a total command of the subject. This expertise has also been apparent in the other topics of the lecture, where he has used, as needed, machine performance observations, numerical analyses, eld trials and centrifuge modelling. As a consequence of this comprehensive approach, conclusions are never commonplace. A constant feature is the consideration and importance given to eld measurements, always the stamp of a good geotechnical engineer. We are thankful to Robert for having drawn our attention to exciting new developments in this area. Ladies and gentlemen, we have had this evening the good fortune to listen to a memorable lecture delivered with the clarity and authority that we have come to expect, as a matter of course, from Professor Mairs presentations. It has been said that it is only possible to transmit experience in the language of science. This lecture is a prime example of the truth of this statement. I am convinced that, well into the future, we shall look back on this occasion as an important landmark that identied the new horizons opening for the perennial and often complex relationship between tunnelling and geotechnics. On behalf of the British Geotechnical Association, I thank you, Robert, for an outstanding lecture, and I call upon all those present here to endorse my thanks by acclamation.

VOTE OF THANKS ANTONIO GENS, Professor of Geotechnical Engineering, Technical University of Catalonia, Barcelona It is an honour, and also a great personal pleasure, to propose the vote of thanks to Professor Mair, the 46th Rankine Lecturer. When Lord Palmerston, the Prime Minister of the day, was invited to the opening of the Metropolitan Line, the rst underground railway in the world, he declined with the argument that, at 79, he wanted to remain above ground as long as possible. We, as a profession, are very fortunate that Robert Mair did not make the same decision at an obviously much younger age. We would have missed an excellent, lucid and well-illustrated lecture that has spanned the whole range of tunnelling issues from stability during construction to the often forgotten long-term settlements that occur long after everybody has packed up and gone home. Robert Mair has proved throughout his career that he is able to move with ease from academic life to professional practice and back while maintaining the same intellectual outlook when tackling both theoretical and practical problems. If there ever was an argument against the possibility of simultaneous excellence in academia and in professional practice, those arguments have been thoroughly disproved tonight, at least in the case of some exceptional individuals. I remember being impressed already by this versatility when