Athens Metro—design and construction of shallow tunnels
Transcription
Athens Metro—design and construction of shallow tunnels
Athens Metro—design and construction of shallow tunnels to control settlements of surface structures By G. Anagnostou, Ch. Kostikas, G. Iakovides & G. Vasilakopoulou RSE Ltd, Consulting Engineers, Zurich & OMETE SA, Consulting Engineers, Athens Tunnelling' 97, 2-4 September, London, The Institution of Mining and Metallurgy Athens Metro—design and construction of shallow tunnels to control settlements of surface structures G. Anagnostou RSE Ltd, Consulting Engineers, Zurich, Switzerland Ch. Kostikas OMETE SA, Consulting Engineers, Athens, Greece G. Iakovides OMETE SA, Consulting Engineers, Athens, Greece G. Vasilakopoulou OMETE SA, Consulting Engineers, Athens, Greece Abstract The paper describes the basic design considerations and the construction of two short shallow double-track tunnels of the Athens metro. In the first tunnel the ground consists of highly fractured peridotites. The minimum distance between the tunnel crown and the foundation level of the building is 2.50 m. The second tunnel crosses decomposed argillaceous rocks (engineering soil) at a depth of 5 m beneath multi-storeyed buildings. In order to avoid damage to the overlying structures, the excavation method and the temporary support must ensure that the settlements will be within tolerable limits. Under the specific project conditions, an approach based upon the observational method is not possible. The performance of the ground is not important because arching cannot relied upon. Since failure could not be allowed to occur, and tolerable deformations during construction would be low prior to the onset of failure, the design has to be less concerned about the variation in rock mass properties and more concerned about the robust structural analysis and detailling of the elements of the support system. In both tunnels, the primary support consists of a reinforced sprayed concrete lining, and tunnel excavation takes place under the protection of an umbrella made by bored steel tubes (forepoling). Besides the short excavation stages and the rapid closure of the ring by a reinforced invert lining, a main feature of the applied method is the systematic reinforcement of the face by fibreglass nails and, in the second tunnel, the full face excavation over a height of 8 m. Furthermore, subvertical micropiles at the footings of the shotcrete shell have been additionally used in the second tunnel. Tunnelling caused in both cases settlements of less than 10 mm. In the first tunnel, the average gross advance rate was 0.80 m/calendar day. Due to the higher amount of protective measures (forepoling, face reinforcement, micropiles), construction progress was considerably lower in the second tunnel (0.35 m/calendar day). So the applied construction methods are very timeconsuming. Nevertheless, they ensure safe execution of the works, lead to minimisation of settlement and are, therefore, appropriate for weak ground tunnelling beneath buildings. 1 Introduction The Athens Metro project comprises the construction of two lines (Line 2 and 3) having a total length of 18 km and 21 stations (Fig. 1). The construction of the Athens Metro started 1991. The major part (11.7 km) of the double-track tunnels is excavated by two TBMs. The remaining 6.3 km are constructed by the cut-and-cover method. This paper focuses on two short tunnels (referred to as tunnel A and B) in which the shotcrete construction method has been applied. Fig. 1 shows the location of these tunnels. The Attiki-Larissa Section of the Line 2 is constructed by the cut-and-cover method except of a 90 m long part (Tunnel A in Fig. 1) where an underground construction method had to be applied due to the surface structures. Construction of Tunnel A was carried out in Spring 1995. The Deligianni-Omonoia Section of Line 2 is excavated by a 9.5 m diameter shield TBM with an opentype cutterhead. At the Karaiskaki Square (B in Fig. 1), the Line 2 passes beneath multi-storeyed buildings. Due to the several tunnel face collapses observed during the TBM operation, a TBM drive beneath the buildings has been considered as too risky. Therefore, the shotcrete construction method was applied for a critical, 25 m long part of the tunnel. The tunnelling works have been carried out in Spring 1996. Both tunnels are in a small distance from the overlying structures. The method of tunnel excavation and support must eliminate the risk of a collapse and minimise the settlements of the building footings. In the paper, the main features of the applied construction methods and the basic design considerations are described. SEPOLIA N TO KIFISSIA LINE 2 ATTIKI A LINE 3 LARISSA AMPELOKIPOI DELIGIANNI OMONOIA B AKADIMIA KERAMEIKOS SINTAGMA LINE 1 TO PIREAS N. KOSMOS LINE 1: EXISTING LINES 2, 3: UNDER CONSTRUCTION DAFNI 1 km Fig. 1 Overview of the Athens Metro and location of the Tunnels A and B 2 Tunnel A Fig. 2 gives an overview of the geotechnical conditions. The tunnel passes beneath an electric substation and a parking building. The distance between the crown of the tunnel and the foundation level of the building is 2.5-4 m. The depth beneath the electric substation is approximately 0.5 m. At the Northern part, the tunnel crosses peridotites. The rock above the crown is locally completely weathered, the upper 3-4 m consist of recent fills. Typical design values are a modulus of elasticity E between 200 and 300 MPa, a cohesion c=20-30 kPa and a friction angle of φ=30°. These values have been established on the basis of field tests and of the observations made during the precedent cut-and-cover works. Some results of pressuremeter tests are shown in the longitudinal section (Fig. 2, borehole ATL103). The geological conditions at the Southern part of the tunnel are more unfavourable (weak rock/ engineering soil with E=75-100 MPa, c=10-20 kPa and φ=25°). The undisturbed ground water table is about at the crown of the tunnel. This is, however, not important for the construction as it was observed that the neighbouring excavations of the cut-and-cover tunnels caused a draw-down of the water level. 5 10 15 m N ATL 102 ATL101 START OF CONSTRUCTION 75 ATL103 ATL100 EXISTING TUNNEL OF LINE 1 PARKING HOUSE +65 ELECTRIC SUBSTATION +60 H-C H M S +50 C 2 C-H 3 +55 E [MPa] Ch. 1+415 0 400 800 Ch. 1+320 M M H H +45 TUNNEL PERIDOTITES 2 3 S M H C MILDLY FRACTURED HIGHLY FRACTURED SLIGHTLY WEATHERED MODERATELY WEATHERED HIGHLY WEATHERED COMPLETELY WEATHERED ENGINEERING SOIL (ALLUVIAL/COLLUVIAL DEPOSITS MODERATELY - COMPLETELY WEATHERED MARL, BRECCIA) Fig. 2 Tunnel A: Site plan and longitudinal section along the tunnel axis (Tunnel cross section: see Fig. 4) 3 Forepoling and stabilisation of the face According to the site investigation results, the rock above the tunnel crown might be completely weathered. In this case, due to the low shear strength of the ground, the concentrated foundation loads acting in a small distance above the crown would lead to the development of slip surfaces (Fig. 3a). A collapse might occur even in the case of fractured rock: in view of the small depth of cover, the possibility of continuous fractures extending from the tunnel crown up to the foundation level cannot be excluded (Fig. 3b). The timely prediction of such a collapse (e.g., on the basis of the observations made during construction) is beyond the Engineer's abilities. Due to these considerations, a forepoling umbrella consisting of 13 m long steel tubes with a diameter of 196 mm has been systematically applied (Fig. 4 and 5). The forepoling minimizes the risk of a collapse for it provides a ground support immediately after each excavation step. Furthermore, the forepoling reduces the settlements of the overlying structures (Fig. 3c) because it offers resistance to the deformations of the ground taking place before the installation of the shotcrete lining. Due to the constraints imposed by the drilling equipment, the tubes form an angle of 5° with the tunnel axis. Placement of the tubes takes place simultaneously with the drilling. Subsequently, the tubes are filled by cement grout. The grout improves the ground locally, i.e. in the immediate vicinity of the tubes. The spacing of the tubes (minimum axial distance 0.40 m) takes into account (i) the stability of the ground in the gap between the tubes and (ii) the loads acting on the tubes. (a) (b) (c) Fig. 3 Risks: (a) Collapse in soil, (b) Collapse due to pre-existing fractures, (c) Settlement ahead of the tunnel face FOREPOLING TUBES 3 1 SHOTCRETE SHELL FIBREGLASS ANCHORS 4 2 5 a a a a Lmin Lmax 2a 2 4m Fig. 4 Tunnel A: Cross section, construction sequence and main elements of the support system 4 Fig. 5 Tunnel A: Installation of the forepoling tubes The forepoling umbrella does not act as an arch. The individual tubes work as beams and transfer the loads to the shotcrete shell and to the ground ahead of the face (Fig. 6a). In order to transfer the loads to the ground, the minimum longitudinal overlapping was chosen to 2 m. Safe support of the forepoling tubes presupposes, furthermore, stability of the tunnel face. A face collapse - even of limited extend - increases the longitudinal span e of the tubes (Fig. 6b). This leads to larger deformations of the tubes and settlements of the overlying structures. To avoid a collapse of the face, the upper tunnel heading has been reinforced by 10 fibreglass friction anchors of 22 mm diameter. The fibreglass can be cutted by the excavation equipment. The minimum length of the anchors (5 m) takes into account the extent of a potential slip line and the necessary anchoring length. During longer interruptions of the excavation works (necessary, e.g., for the installation of the forepoling tubes), a shotcrete layer was applied to the face. Due to the plane form of the face, the statical function of this layer is limited to the protection from local peel-offs. Excavation Method The cross section has been divided into three parts (Fig. 4). Due to the forepoling (which provides ground support immediately after each excavation step) and to the application of fibreglass anchors (which eliminates the problem of face stability), a further division of the cross section is not necessary. Criteria for the height of the three parts were (i) the stability of the respective faces and (ii) the size of the available excavation equipment. The excavation step a (=1-1.30 m) and the distance L (=5-8 m) between the upper face and the invert are the main parameters of the excavation method. As explained below, the determination of these parameters must take into account contradictory criteria. 5 e e+Äe a 2 (a) 4m (b) Fig. 6 Free span e of the forepoling tubes (a) under normal conditions, and (b) in the case of face instability 2 (a) 4m (b) Fig. 7 Possible consequences when closing the ring (a) far away from the tunnel face, (b) very close to the tunnel face 40 u p 30 1 2 3m 0 15 SETTLEMENT u [mm] FOUNDATION LEVEL E=150 MPa 50 20 25 c=20 kPa c=60 kPa 0 10 500 0 0 p cr 50% 100% SUPPORT PRESSURE p [% OF INITIAL STRESS ] Fig. 8 Ground response curves (Inset: discretisation and plastic zone for p=p cr ) 6 A closed ring offers greater resistance to the deformations of the surrounding ground than an "open" arch (Fig. 7a). In particular, the footings of the arch represent known weak points. In practice, the contact between the shotcrete shell and the underlying ground at the footing is often not perfect. The ground starts to bear only after some settlement of the shell has taken place. The rapid formation of a closed ring (within a diameter from the excavation face) is, therefore, essential for the reduction of the settlements (Weber, 1980). On the other hand, a very small distance L means a steeply inclined face, and is therefore unfavourable for its stability (Fig. 7b). With limit equilibrium calculations, it was estimated that an L-value smaller than 5 m would make necessary the application of fibreglass anchors over the entire face (top heading and bench). Similar considerations apply to the determination of the round length a. A large round length a is favourable concerning the net excavation rate. On the other hand, a larger round length leads to larger deformations of the forepoling tubes as well as to higher loads acting on the fresh shotcrete arch and the core ahead of the face (Fig. 6). Note that the deformations of the tubes increase with the fourth power of their span: by increasing the step a from 1.00 m to 1.50 m, the deformations become higher by a factor of 5. So, the larger the excavation step, the stronger the shotcrete shell, the heavier the forepoling and the denser the face-reinforcement must be. The advantage of a higher net excavation rate is, therefore, balanced out by the time needed to install the heavier support. Design calculations Elements of a robust design are the formation of a clear picture of the statical system and of the flow of the forces, the recognition and evaluation of potential mechanisms leading to collapse or inadmissible settlement, the evaluation of the countermeasures and the structural detailing. These works cannot be substituted by statical calculations. The computations provide, however, valuable indications of the structural behaviour and are therefore indispensable in the decision making process (Kovári, 1972). In this respect, the computational results of Fig. 8 are interesting for they provide some insight to the general behaviour of a shallow tunnel in weak ground. The diagram has been obtained by means of elasto-plastic calculations. It shows the relation between the support pressure p and the settlement u at the foundation level. The support pressure is expressed as a percentage of the initial stress. The curves hold for different values of the modulus of elasticity E and of the cohesion c of the ground. The smaller the support pressure p, the higher the settlement u will be. When the support pressure approaches a critical value pcr, the settlement becomes asymptotically infinite. The system has reached limit equilibrium state. The plastic zone extends from the tunnel boundary up to the foundation level (hatched area in the inset of Fig. 8). A further reduction of the support pressure leads to collapse. Why is this example interesting ? As shown by the diagram, the critical support pressure depends on the ground cohesion c, but not on the modulus of elasticity E (see for example the curves for c=60 kPa and E=150-500 MPa). Consequently, the settlement observed up to the beginning of collapse may be very small. In view of the small deformations, the site engineer might take into consideration a reduction of the support measures, whereas the structure is close to limit equilibrium. Deformation measurements provide, unfortunately, little information on the safety factor. Under the specific project conditions, an approach based upon the observational method is not possible. The performance of the ground is not important because arching cannot relied upon. In the design calculations, the loads arising from the overburden and the buildings must be taken into account without any reduction. Since failure could not be allowed to occur, and tolerable deformations during construction would be low prior to the onset of failure, the design has to be less concerned about the variation in rock mass properties and more concerned about the robust 7 structural analysis and detailling of the elements of the support system. In the present case, the calculations concerned the deformations and the bearing capacity of the forepoling and of the shotcrete shell, the stability of the excavation faces and the settlement of the overlying structures. The bearing capacity assessment of the forepoling tubes and of the shotcrete shell has been carried out according to the Eurocodes 2 and 3. The stresses in the shotcrete shell and the settlements have been estimated by numerical analyses with the finite element method. The statical system consists of the shell and the surrounding ground up to the foundation level of the building. In the used computer codes TUNNEL (RIB, 1994) and HYDMEC (Anagnostou, 1991), the ground is modelled as an elasto-plastic material. However, for reasons explained above, plastic behaviour is of minor importance. The bearing capacity of the shotcrete shell has been checked also with the more unfavourable computational model of loosening pressure. In these calculations, the ground above the tunnel crown does not belong to the statical system, but it is taken into account only as a load. Finally, three-dimensional shell calculations have been carried out with the computer code SCADA (1987) in order to analyse the stresses which develop in the incomplete shotcrete lining due to the loads from the forepoling (Fig. 6a). 1 1 Fig. 9 1-1 2 4m Examined collapse mechanisms The forepoling tubes have been considered as beams loaded by the overlying ground, elastically supported by the shotcrete shell and elasto-plastically supported by the ground ahead of the face. The elasto-plastic support takes into account the stiffness and the bearing capacity of the fibreglassreinforced core. The stability of the tunnel face was examined with three-dimensional limit equilibrium calculations, which take into account the loads imposed by the forepoling tubes. Similar calculations have been carried-out for the stability of the bench and of the lateral walls of the tunnel (Fig. 9). The Tunnel construction The tunnel excavation started in the May 1995 and was completed without any problems after 4 months (Fig. 10). The average gross production rate was 0.80 m/calendar day. This figure includes the interruptions for the installation of the forepoling (4-5 days per group of 23 tubes). The tunnelling-induced settlements have been small (Fig. 11). The maximum settlement has been observed at the electric substation (10 mm). The settlement of electric substation developed slowly over a period of 3-4 months, and, furthermore, it does not correlate with the tunnel crown settlement. A plausible explanation for this observation was not found. The columns of the building experienced settlements up to 5 mm. In the Fig. 12, the settlement of the columns A and C is plotted over the tunnel face position x. According to the diagram, settlements start to occur approximately 5 m ahead of the face. The settlements reach their final value within a distance of approximately 10 m from the face, i.e. with the construction of the invert lining. 8 [m] 80 60 40 20 0 0 30 y /da m 0.8 60 90 120 days Fig. 10 Tunnel A: Progress of the works 1 A C SECTION 1-1 D E F G K L 1 0 5 10 mm Fig. 11 Tunnel A: Final settlements x ä -30 -20 -10 0 EXCAVATION 10 20 x[m] 2 A C 4 6 ä [mm] Fig. 12 Tunnel A: Development of the settlements of the columns A and C 9 Tunnel B Due to the positive experiences from the tunnel construction in Attiki Square, it was decided to apply the same method for a critical, 25 m long part of the Deligianni-Omonoia Tunnel beneath multi-storeyed buildings. Up to a depth of approximately 4 m, the ground consists of alluvium deposits. The underlying rock is completely weathered ("engineering soil") with an erratic weathering profile. Based upon back-analyses of the observations made during the precedent TBM drive, the following design values have been established: E=50 MPa, c=10 kPa and φ=28°. The Fig. 13 shows the excavation method. The size and the form of the tunnel takes into account the need to provide space to allow the TBM to pass through. Excavation started from a shaft. In a first construction phase (referred to as "Phase I" in Fig. 13), parts 1 and 2 of the cross section have been excavated in alternating steps of 1 m. The shotcrete shell is 0.35 m thick and the mesh reinforcement amounts to 9.40 cm2 per linear metre. The reinforcement overlapping amounts to minimum 0.50 m in the cross section and 0.30 m in the longitudinal direction. Excavation of part 2 follows the heading of part 1 in a distance of 3.60 m, thus making possible the rapid closing of the 0.25 m thick temporary invert shell. The invert lining is reinforced by 5.88 cm2 per linear metre. Steel sets have not be used because the forepoling umbrella provides protection immediately after each excavation step and the contribution of the steel sets to the stiffness of the temporary structure is very low relative to the one of the shotcrete. SHAFT 5 10 15 m FOUNDATION LEVEL FOREPOLING SHOTCRETE SHELL 1 FIBREGLASS 2 MICROPILES TEMPORARY INVERT CONSTRUCTION PHASE I 3 CONSTRUCTION PHASE II Fig. 13 Tunnel B: Excavation method 10 Fig. 14 Tunnel B: Installation of the subvertical micropiles close to the face Construction-phase II started after completion of phase I. In this phase, the temporary invert lining has been stepwise removed, part 3 of the cross-section has been excavated and the reinforced invert lining of part 3 was shotcreted. After completion of phase II, the shotcrete lining forms an almost circular ring. The shotcrete lining of phase I is obviously less favourable than the one of phase II. Concerning the settlement of the overlying buildings, construction phase I is more critical than the phase II. The construction in two phases was, however, necessary due to the size of the available equipment. In order to reduce the settlements in the critical phase I, alternative solutions have been examined. The settlement can be reduced either by means of a temporary invert lining or by means of micropiles at the footings of the shell. Due to the variability and the erratic character of the ground properties, large uncertainties existed concerning the main design parameter for the micropiles, i.e. the bond stress between the pile and the surrounding ground. Furthermore, the bearing capacity of the micropiles is low for subhorizontal loads. An invert lining is more efficient for preventing horizontal displacements. On the other hand, the connection of the invert lining to the upper shell is a structurally weak point. Due to the low stiffness of the ground, the shear force is very large. In spite of the remarkable curvature of the invert lining, a very high amount of stirrup reinforcement would be necessary for safe transmission of the shear force. For these reasons, it has been considered as too risky to rely solely upon the micropiles or solely upon the invert lining. Under the given geotechnical situation, it is rather the combined action of these two elements which ensures that the settlements will be within the tolerable limits. This has been verified also by finite element computations which take into account the deformation characteristics of the micropiles, the stiffness of the shotcrete shell and the stiffness of the ground. The spacing of the micropiles is equal to the round length, i.e. 1 meter. The micropiles are 8 m long. Installation takes place after each excavation step immediately at the face (Fig. 14). The inclination of the micropiles is 30° from the vertical. Each micropile consists of a steel tube with a 11 diameter of 250 mm in which a second tube with a diameter of 200 mm is inserted. The tubes have a wall thickness of 6.3 mm. The base of the shotcrete shell has been additionally reinforced in order ensure a safe load transfer from the micropile to the shotcrete. Due to the higher loads, the forepoling umbrella is heavier than in Tunnel A: at the crown (over a sector of 90°) the tubes are installed side by side. Furthermore, to avoid collapse of the lateral excavation faces, the forepoling extends over the entire circumference of part 1. The axial spacing of the lateral tubes amounts to 0.50 m. Each forepoling pile consists of two tubes having a diameter of 193.7 mm and 168.3 mm, and a wall thickness of 6.5 mm and 5.6 mm, respectively. The tubes are 12 m long with an overlapping of 6 m, i.e. in each cross section two rows of tubes are present. To increase the bearing capacity of the ground ahead of the 8 m high face, the core was reinforced by 58 horizontal micropiles having a minimum length of 6 m (i.e., 1 pile/m2). Due to the engineering soil conditions prevailing in this tunnel, the main design parameter for the face reinforcement was not the tensile strength of the fibreglass bar, but rather the bond stress between pile and ground. The low bond stress (100 kPa) made necessary a micropile diameter of 120 mm. The tunnel was constructed during Spring 1996. The settlements have been between 2 and 5 mm. Due to the higher amount of protective measures (forepoling, face reinforcement, micropiles), the gross advance rate was only 0.35 m/calendar day and thus considerably lower than the one of the Attiki Sq. tunnel. For the 25 m tunnel, approximately 70 days have been necessary. Conclusions The applied construction methods are time-consuming. Nevertheless, they ensure safe execution of the works, lead to minimisation of settlement and are, therefore, appropriate for weak ground tunnelling beneath buildings. References 1. Kovári K.: Design Methods for Underground Structures. Int. Symp. f. Untertagbau, 1972, pp. 198-223. 2. Weber J.: Konstruiren, rechnen und messen. U-Bahn-Linie 8/1 (Ed.: Firmengruppe & Referat München), 1980, pp. 119-130. 3. RIB Bausoftware GmbH: TUNNEL. Grundbau- und Tunnelstatik mit FEM. User's 1994. 4. Anagnostou G.: Untersuchungen zur Statik des Tunnelbaus in quellfähigem Dissertation 9553, Swiss Federal Institute of Technology, Zurich, 1991. 5. SCADA Systems Corp.: SCADA User's Manual, 1987. Lucerne, U-BahnManual, Gebirge. 12