Engineering geology
geological conditions

Full-face excavation of large tunnels in difficult conditions

Introduction

No established method was available at the beginning of the 20th century for the design and excavation of tunnels in difficult conditions. However, with the need of new transportation lines and the increasing mobility, spectacular tunnels were excavated and completed. Important examples in Europe are the Frejus and Simplon railway tunnels through the Alps and “La Grande Galleria dell’Appennino” through the Apennines in Italy, as shown in Fig. 1. Different tunneling methods were applied such as the Belgian, Austrian and Italian methods that were characterized by significant differences in the choice of the section where to initiate excavation. In general, tunneling used to take place in stages by the sequential excavation method.

Fig. 1. “La Grande Galleria dell’Appennino”, excavated through the Apennines along the Bologna-Florence railway line between 1921 and 1934.

The need to improve the way of thinking of the time was soon recognized in order to make tunneling a more systematic and efficient work. Simplon Engineers Andre and Rothpletz emphasized the advantage of full-face excavation already one hundred years ago (e.g. Andreae, 1956). Rabcewicz (1964) pointed out that tunnels should be driven full-face whenever possible. It is indeed in this context that in the mid-1980s, Lunardi (2000) saw the importance of the stability of the face, in particular with the increase of the size of tunnels and their depth below surface, and promoted the full-face method. He suggested that understanding and controlling the behavior of the “core” ahead of the advancing tunnel face are the secret to successful tunneling in difficult conditions.

Tunnel construction in such conditions (e.g. in poor quality ground or in squeezing or unstable ground) is very demanding and reliable predictions at the design stage are difficult if not impossible, so that most often the use of the “interactive observational approach” is advocated. If consideration is given to the construction of deep tunnels (such as the new Alpine Tunnels in Europe, e.g. Lötschberg, Gotthard, Lyon-Turin, and Brenner Base Tunnels), alignment constraints, and uncertainties of geological exploration, it is not always possible to avoid difficult conditions. Therefore, the selection of the most appropriate excavation-construction method is highly problematic and uncertain. The choice is in all cases between mechanized tunneling (tunnel boring machine, TBM) and conventional tunneling.

In mechanized tunneling, due to the fixed geometry and the limited flexibility of the TBM, allowable space to accommodate ground deformations is restricted. On the contrary, in conventional tunneling where a considerably larger profile can be excavated in order to allow for large deformations, inevitable excavation will take place with a low rate of advance. It is however true that, if the work is well planned and appropriate stabilization measures are implemented, excavation may proceed at an acceptable rate of advance even in very difficult ground conditions, as in squeezing conditions.

The purpose of this paper is to describe the full-face excavation method, which has been experienced successfully so far under various conditions, at depth or near the ground surface. In doing this, the attention is paid to the use of the so-called “heavy” and “light” methods of tunneling, meaning that limited or significant deformations of the ground around the tunnel are allowed to take place in a controlled manner. An attempt is made through a case study to outline some of the geotechnical issues in view of modeling the tunnel behavior and observation and monitoring during excavation.

Full-face tunneling

Full-face tunneling has been applied in Italy for more than 30 years, with cross-sectional areas ranging from 120 m2 to 220 m2, in different ground conditions, near the surface or at depth (Lunardi, 2000). Fig. 2 and Table 1, taken from Lunardi and Barla (2014), show a selection of these tunnels with the purpose of conveying the wide spectrum of ground conditions where the method has been applied successfully. In addition to major railway, highway and road tunnels in Italy, also reported in Fig. 2 is the “Tartaiguille” tunnel in France, excavated along the railway line “TGV Mediterranée”.

Fig. 2. Photographs showing the face of a few selected tunnels excavated full-face by using the “heavy method” (Lunardi and Barla, 2014).
Table 1. Selected projects (see Fig. 2).

Based on the wide experiences gained so far, the most important component of successful design and construction of tunnels when full-face excavation is considered is the ability to understand the importance of the ground deformational response. This is strictly linked to the formation of the “arching effect” (i.e. the ability of the ground to stabilize and sustain itself), needed for reaching stability of the underground excavation, in the short and long terms.

As illustrated in Fig. 3, the designer has to make appropriate predictions of the “extrusion” (longitudinal displacement) of the face, the “pre-convergence” (radial displacement ahead of the face) and the “convergence” of the tunnel perimeter (radial displacement behind the face), before starting excavation. From this point of view, one essential component of the approach is the full understanding of the tunnel behavior in three-dimensional conditions.

Fig. 3. Simplified illustration of the ground deformational response during full-face tunneling.

When full-face tunnel excavation is carried out, stresses and strains do develop in the surrounding rocks, including the “core” ahead of the advancing tunnel face. This “core” is the main tool for reaching the tunnel stability by means of stabilization and reinforcement measures as appropriate. The full-face method, as applied for the first time in Italy (Lunardi, 2000), is indeed characterized for having developed the technologies for “protecting” and “reinforcing” this “core” ahead of the advancing tunnel face, whenever needed (in particular with the use, in weak rock, of fiber-glass elements).

The transition from the confinement action due to the “core” ahead of the advancing tunnel face to that of the support along the tunnel perimeter is to take place in the most uniform and gradual way as possible, by placing the invert in the near vicinity of the face, when needed. The key component of the approach is to minimize the extrusion surface, which coincides with the tunnel perimeter and extends longitudinally, from the point of contact between the ground and the support at the crown and at the invert, respectively.

“Heavy” and “light” methods in full-face tunneling

In general, the major problems encountered when tunneling in squeezing rock, which is a typical difficult ground condition, are associated with the stability of the tunnel and of the face. As illustrated in Fig. 4, obtained by the author with conventional axisymmetric and plane strain models by the finite element method (FEM), a plastic zone develops in the rock mass surrounding the advancing tunnel. Depending on the rock mass properties, the wall plastic zone may interact or not with the face plastic zone.

Fig. 4. Plastic zone developing around a 10 m diameter circular tunnel (Left: axisymmetric, right: plane strain FEM analyses; in-situ stress p0 = 5.0 MPa, deformation modulus Ed = 400 MPa, rock mass strength σcm = 0.6 MPa, dilation angle ψ = 0.0°).

As already noted, of significant importance for the understanding of the tunnel response are both the radial displacements of the tunnel wall and the corresponding longitudinal displacements of the tunnel face as excavation proceeds. The tunnel face follows the same deformational pattern as the tunnel itself, although the longitudinal displacements of the “core” ahead of the face are significantly smaller than the tunnel radial displacements. As shown by Hoek (2001), this is well illustrated in squeezing rock by plotting the normalized wall convergence (ɛt) and normalized axial displacement of the tunnel face (ɛf) against the ratio of rock mass strength (σcm) to in-situ stress (p0), i.e. σcm/p0. Note that ɛt is defined as the percentage ratio of radial tunnel wall displacement ur to tunnel radius a and ɛf as the percentage ratio of axial face displacement uf to tunnel radius a.

Fig. 5 shows such a plot, obtained for a 10 m diameter circular tunnel at depth where the rock mass is represented as a continuum, isotropic, elastic-perfectly plastic model with a Mohr-Coulomb yield surface and by assuming a zero dilation. The results were generated with an axisymmetric FEM model under no support pressure either at the wall or at the face. It is shown that the strains ɛt and ɛf increase asymptotically when the ratio of the rock mass strength to in-situ stress is smaller than 0.2, to indicate the onset of severe instability of the tunnel if no adequate support measures are implemented (Hoek, 2001).

Fig. 5. Normalized wall convergence (ɛt) and normalized axial displacement of the tunnel face (ɛf) versus the ratio of rock mass strength (σcm) to in-situ stress (p0), i.e. σcm/p0.

Of the available options for conventional tunnel excavation (e.g. multiple headings, top heading and benching down, full-face), the choice falls, also in difficult conditions such as in squeezing rock, on the full-face method (Fig. 6) with a systematic reinforcement of the ground ahead of the face, e.g. by means of fiber-glass elements. However, this method, well-known from several tunnels mostly excavated in Italy (Fig. 2 and Table 1), including tunnels in difficult conditions, was applied in combination with a stiff support. This was in general possible and applied successfully due to a rather moderate overburden and when squeezing was not very severe.

Fig. 6. Full-face excavation and construction method: face reinforcement and ring closure: “heavy method” (left) and “light method” (right).

With the need to cope with high overburden and in particular very severe squeezing conditions as, for example, in the Sedrun section of the Gotthard Base Tunnel in Switzerland, the proposal was made (Kovári and Ehrbar, 2008) to combine full-face excavation with a yielding support system composed of steel sets with sliding connections. This challenging combination raised questions on kinematics, stability and handling of the steel sets. Advanced structural analyses and full-scale field tests were carried out and the tunnel was completed successfully.

As illustrated in Fig. 6, with this combination of structural elements, one is moving (Kovári, 1998) from the “heavy method” (“resistance principle”), where only small deformations are permitted, to the “light method” (“yielding principle”), where large deformations are allowed to develop around the tunnel with the expectation that rock pressure will decrease with increasing deformations.

With the “heavy method”, the primary lining is designed to be very stiff (generally composed of steel-fiber shotcrete and heavy steel sets). The tunnel cross section is entirely open and the primary lining is installed near to the face (in Fig. 6 left, which is for a typical tunnel in Italy; the “ring is closed quickly” by using a steel set as invert). The final concrete invert (first) and final concrete lining (second) are cast within a short distance from the face. It is apparent that if very high rock pressures are expected, as in deep tunnels, this solution soon becomes impractical.

With the “light method”, the excavation profile is chosen in order to maintain the desired clearance and to avoid the need for re-profiling. A key point is to be able to control the development of deformations. A suitable tunnel support system is to be adopted (in Fig. 6 right, which is referred to the Gotthard Base Tunnel, a yielding support by steel sets with sliding connections was used, including steel bolts for face support) that will allow for accommodating deformations without damage of the lining.

The face stability when driving a tunnel consists in reinforcing the rock mass ahead by means of grouted fiber-glass elements. There are a number of fiber-glass elements that may be adopted. Both smooth and corrugated tubes are available. More recently, flat elements (Fig. 7) are being used which can be assembled in-situ in a wide variety of types; they are very easy to inject and transport, and they allow reinforcement advance steps up to 25 m. In typical reinforcement schemes, the fiber-glass elements are used to reinforce the “core” ahead of the face and in cases to provide a “reinforced ring” around the tunnel.

Fig. 7. Flat fiber-glass structural elements adopted for face reinforcement in the full-face excavation and construction method.

There are a number of options for the yielding support, in addition to providing sliding joints in the top hat steel sets as in the case of the Gotthard Base Tunnel in the Sedrun section. These include the LSC (lining stress controller) element (Schubert, 1996), the WABE honeycomb element (Moritz, 2011), and the hiDCon element (Kovári, 2008). Fig. 8 illustrates typical stress-strain characteristics of these elements. As shown in this figure, a yielding element initiates yielding at a specified stress level as it undergoes a very small strain (smaller than 5% approximately). Deformation continues with the element length shortening significantly before the stress in the element increases.

Fig. 8. Stress-strain characteristics of LSC (lining stress controller), WABE and hiDCon elements obtained in laboratory tests.

Full-face tunneling in severely squeezing conditions: a case study

In order to illustrate full-face tunneling in severely squeezing conditions, a case study is discussed in the following. Reference is made to the Saint Martin La Porte access adit, along the Lyon-Turin Base Tunnel, where a yield-control support system combined with full-face excavation has been adopted successfully in order to cope with the large deformations experienced during face advance through the Carboniferous formation (see Barla, 2009, Barla et al., 2010).

Rock mass conditions

The rock mass encountered during excavation, as shown in Fig. 9, is a highly heterogeneous, disrupted and fractured rock mass, which is often affected by faulting that results in a significant degradation of the rock mass conditions. The overburden along the tunnel in the zone of interest ranges from 300 m to 600 m. Excavation takes place in dry conditions.

Fig. 9. Typical geological conditions at the tunnel face (gps – sandstone; a – clay-shales; c – coal, etc.).

Support system

The design concept consists in the systematic use of full-face excavation and reinforcement, coupled with a yield-control support system by using a near circular cross section (radius of 6.10 m), as shown in Fig. 10, Fig. 11. The excavation-support sequence can be summarized as follows:
(1) Stage 0: face reinforcement, including a ring of fiber-glass elements around the tunnel perimeter, over a 2–3 m thickness.

(2) Stage 1: mechanical excavation carried out in steps of 1 m in length, installation of 8 m long rock dowels along the perimeter, yielding steel sets with sliding joints, and a 10 cm thick shotcrete layer. The tunnel is excavated in the upper cross section to allow for a maximum convergence of 600 mm.

(3) Stage 2: the tunnel is opened to the full section at a 30 m distance from the face, with the application of 20 cm shotcrete lining, yielding steel sets with sliding joints fitted with hiDCon elements. The tunnel is allowed to deform in a controlled manner with a maximum convergence not to exceed 400 mm.

(4) Stage 3: installation of the final concrete lining at a distance of 80 m from the face.

Fig. 10. Yield-control support system. Near circular cross section (R = 6.3 m). Stages 1 and 2 are shown. Face and ring reinforcement are not indicated.
Fig. 11. Left: view of the tunnel during excavation. Right: the hiDCon elements as installed.

As shown in Fig. 10, Fig. 11, the most important component of such a yielding support is the hiDCon element. Nine such elements (one in the invert) are installed in slots in the shotcrete lining between the yielding type steel sets. These elements (height of 40 cm, length of 80 cm, and thickness of 20 cm) yield at approximately 40–50% strain with a yield stress of 8.5 MPa. With 9 elements installed, if each element may attain a 50% strain, the maximum allowed radial displacement ΔR is equal to 20 cm approximately, resulting in a total tunnel convergence of 40 cm. Also, if one takes a yield stress of 8.5 MPa, the radial confinement stress on the surrounding rock results to be 0.3 MPa approximately.

Performance monitoring

Monitoring of tunnel convergence has been underway along the tunnel systematically. Convergences were measured by means of optical targets placed along the tunnel perimeter. Also measured were the longitudinal displacements ahead of the face. In addition, a number of sections have been equipped with multi-position borehole extensometers and strain/stress meters in the primary and final linings. A few representative monitoring data are described as follows.

Fig. 12 shows the convergences measured in Stage 1, along different arrays (1–5, 1–3, 3–5) between chainage 1100 m and 2300 m approximately, with the tunnel face 15 m ahead of the monitoring section. Similarly, Fig. 13 gives the convergence at 30, 80 and 120 days following the opening of the full cross section in Stage 2.

Fig. 12. Tunnel convergence versus chainage in Stage 1 along different arrays.
Fig. 13. Tunnel convergence versus chainage in Stage 2 along array 1–5 at 30, 80 and 120 days, following the opening of the full cross section.

One should note that the new yielding support system shown in Fig. 10, Fig. 11 was applied systematically only following chainage 1350 m approximately. Prior to this chainage, a horseshoe tunnel section was excavated full-face with fiber-glass reinforcement. The support consisted of steel sets with yielding connections and shotcrete. Starting with chainage 1550 m, the ground conditions were improved substantially. In the sections with the new yielding support system installed, the mean tunnel normalized convergence 15 m behind the face in Stage 1 is 4% and locally never in excess of 6–7%. In Stage 2, the maximum tunnel normalized convergence is 5% around chainage 1480 m, in excess with respect to the target value of 3.5%.

Multi-position borehole extensometers were also used in order to observe the rock mass response in the ground around the tunnel. Fig. 14, Fig. 15 show typical monitoring results for the section at chainage 1444 m. It is of interest to point out the significantly non-symmetric closure of the tunnel, due to the essentially anisotropic features of the rock mass and the presence of “strong” (sandstones and schists) and “weak” (coal and clay-like shales) “layers” which dip from the left to the right of the tunnel cross section.

Fig. 14. Radial displacement versus distance at different times around the tunnel at chainage 1444 m.
Fig. 15. Zone around the tunnel at chainage 1444 m, where the measured radial strain is greater than 1%, obtained by using multipoint extensometers.

Fig. 16 illustrates the maximum tangential stress in the final lining, at the sidewalls, versus time. In general, in the cross sections following chainage 1350 m, with the new yield-control support system installed, this stress is between 2 MPa and 10 MPa, with an average value equal to 7 MPa approximately. The corresponding stress in the final lining, where however excavation took place with the previous support formed with steel sets with yielding connections and shotcrete, at chainages 1229 m and 1323 m, is 22 MPa and 27 MPa, respectively. Also noted, as shown in Fig. 15, is the longer transient phase with a more pronounced stress rate value.

Fig. 16. Variation of tangential stress with time in the final lining at different chainages, obtained by using embedded strain meters.

Numerical modeling

Significant features of the tunnel response are the time-dependent deformations observed whenever face advance is stopped. Also, these time-dependent deformations do take place during excavation, when it is difficult to distinguish the “face effect” from the “time effect”. In addition, laboratory tests on representative specimens show a time-dependent behavior (Debernardi, 2008, Debernardi and Barla, 2009). Therefore, in such condition modeling should consider the use of constitutive laws that account for time-dependent behavior.

Many constitutive laws can be used to describe such a behavior; however, only few of them can reproduce satisfactorily all the time-dependent features involved in tunnel excavation, with a reasonably simple mathematical formulation to be used in design practice. With the phenomena observed in the Saint Martin La Porte access adit in mind, the SHELVIP (Stress Hardening ELasto VIscous Plastic) model was formulated (Debernardi, 2008, Debernardi and Barla, 2009). This model is used as follows.

With the objective to analyze the tunnel response based on the available monitoring data previously described, the axisymmetric finite difference model shown in Fig. 17 has been created with the FLAC computer code (Itasca, 2006). The tunnel cross section in the model is circular, with an equivalent radius of 6 m. The total size of the model (96–280 m) is such as to minimize boundary effects. With the overburden equal to 360 m, the in-situ stress is 9.8 MPa and isotropic. It is understood that the model created oversimplifies the observed conditions during excavation, in view of the significantly non-symmetric closure of the tunnel as shown in Fig. 15.

Fig. 17. The axisymmetric finite difference model.

As shown in Fig. 17, an equivalent pressure of 0.1 MPa applied to the face represents the ground reinforcement ahead of the face. An equivalent radial pressure equal to 0.1 MPa, 5 m behind the tunnel face, represents the radial reinforcement and the first stage lining (10 cm thick). An additional internal pressure 30 m behind the face, which reaches the constant value of 0.383 MPa in a 5 m span, simulates the influence of the second stage lining. This value accounts for the yielding strength of the yielding elements, as known from laboratory tests. The assumed excavation rate is 0.54 m/d.

Back-analysis carried out with the model has led to the values of the constitutive parameters listed in Table 2. Fig. 18 shows the comparison of computed and measured radial displacements for the tunnel section at chainage 1444 m. The numerical results with the mean curve agree satisfactorily with the monitoring data, notwithstanding the scattering due to the high heterogeneity and anisotropy of the rock mass. Also the displacements around the tunnel monitored with the multi-position borehole extensometers are satisfactorily reproduced, as shown in Fig. 19.

Table 2. Parameters for the SHELVIP model, from back-analysis of the monitoring data.
Fig. 18. Computed versus monitored convergences at chainage 1444 m.
Fig. 19. Computed versus monitored radial displacements around the tunnel at stages 1 (a) and 2 (b), chainage 1444 m.

Conclusions

This paper presented the state-of-the-art in full-face tunneling in difficult conditions, when the tunnel stability relies heavily on the ground reinforcement ahead of the tunnel face (the so-called “core”) and, in cases, on a ring around the tunnel. The developments of the method, with the primary lining designed to be very stiff (generally composed of steel fiber shotcrete and heavy steel sets), as initially applied mostly in Italy, were mentioned. Then, the early applications of full-face tunneling under difficult conditions by using a yielding support system were described.

The well-known case study of the Saint Martin La Porte access adit, along the Lyon-Turin Base Tunnel, which experienced severely squeezing conditions during excavation in the Carboniferous formation, was taken as representative. In this tunnel, which has been widely studied by the author, an innovative tunnel excavation and construction method was introduced which combines full-face excavation and face reinforcement by means of fiber-glass elements with a yield-control support. The results of in-situ performance monitoring and numerical modeling with consideration of the rock mass time-dependent behavior were illustrated.

Source: Full-face excavation of large tunnels in difficult conditions

Author: Giovanni Barla

One thought on “Full-face excavation of large tunnels in difficult conditions

Leave a Reply

%d bloggers like this: