Since the last decades of the 20th century, there has been a rapid development in various stages of geotechnical design, analysis and computational methods. Yet, regardless of the capabilities offered by the numerical tools, the results can still involve uncertainties when parameters are used directly without considering the actual failure mechanism of the rock mass in tunnelling. Understanding the rock mass behaviours in tunnelling can ensure selecting appropriate design parameters (for rock mass and/or discontinuities) and failure criteria to be used in numerical analysis and consideration of the principles in association with tunnel support.
Engineers can design reinforced concrete or steel structures using certain checks for specifically predefined failure mechanism. Specifically, design should consider bending moment, axial force, shear, penetration and deflection (serviceability limit state). In tunnelling, however, there is no specific procedure to check against a predefined failure mechanism. This paper points out that the first step is not to start performing numerous calculations (probably misleading or useless), but to define what the potential failure mechanisms are and to qualitatively consider the support theories to account for them. This process is thus applied for the heterogeneous rock masses of flysch (Fortsakis, 2014).
Rock mass behaviour evaluation in tunnelling and its relation with the design process have been significantly reported. Goricki et al., 2004, Schubert, 2004, Potsch et al., 2004 and Poschl and Kleberger (2004) have studied rock mass behaviours with respect to design and construction experiences of Alpine tunnels and Palmstrom and Stille (2007) from other tunnels. Flysch rock is composed of varying alternations of clastic sediments associated with orogenesis, since it ends the cycle of sedimentation before the paroxysm folding process. Intense folding and heavy shearing with numerous overthrusts thus characterise the environment in areas of flysch formations. It is characterised mainly by rhythmic alternations of sandstone and pelitic layers (siltstones, silty or clayey shales), where the thickness of sandstone or siltstone beds ranges from centimetres to metres. Consequently, conglomerate beds may also be included. The main thrust movement is associated with smaller reverse faults within the thrust body. The overall rock mass is highly heterogeneous and anisotropic, and thus may be affected by extensional faulting producing mylonites. The tectonic deformation drastically degrades the quality of the rock mass, a reason that flysch is characterised by diverse heterogeneity (Fig. 1) and the presence of low strength and tectonically disturbed structures (Fig. 2). Such formations are classified into 11 rock mass types (I–XI) according to the siltstone–sandstone proportion and their tectonic disturbance.
The design of tunnels in weak rock masses such as disturbed and sheared flysch presents a major challenge to geologists and engineers. The complex structure of these materials, resultant from their depositional and tectonic history, means that they cannot easily be classified in terms of the commonly used characterisation schemes.
The variety of geological conditions under different in situ stresses, in both mild and heavy tectonism examined here, provided significant amount of information regarding the engineering geological conditions and geotechnical behaviour of several flysch rock mass types. These behaviours were analysed and evaluated so as to define the geotechnical characteristics for each flysch type.
This study is based on experiences obtained from the design and construction of 62 mountainous twin tunnels of the Egnatia Highway in Northern Greece. The cross-section of these tunnels is 100–120 m2, constructed conventionally using the top heading and bench method. In this context, a database named “Tunnel Information and Analysis System” (TIAS) was created (Marinos, 2007, Marinos et al., 2013). Using this database, the evaluation of huge geological and geotechnical data from the design and the construction of 12 tunnels is presented. These cases comprise tunnelling up to 500 m of overburden depth.
The data processed by TIAS are obtained from geological mapping (design and face mapping records), boreholes, laboratory tests, site testing, geotechnical classifications (design and construction records) and designation of design parameters. Data were also collected and processed in view of the geotechnical behaviour, such as deformations, overbreak, structural failures and groundwater inflow. Data from detailed information on temporary support measures and tunnel construction cost were also included. The processing and evaluation of this information contributed to assessing the correlations between behaviours of the ground and the formulation and the temporary support requirements. The use of TIAS database enabled then the determination of the possible rock mass types of flysch and the engineering geological characterisation in terms of properties and their behaviour in underground construction (Marinos et al., 2013).
The development of powerful microcomputers and of user-friendly software prompted a demand on data related to rock mass properties required as inputs for numerical analysis or close-form solutions for designing tunnels. This necessity preceded the development of a different set of rock mass classifications, where the geological strength index (GSI) is such a classification. The Hoek–Brown failure criterion (Hoek et al., 2002) is closely connected to the GSI, covering a wide range of geological conditions affecting the quality of the rock masses, including heavily sheared weak rock masses (Hoek et al., 1998). The GSI considered as such a tool for assessment was initially introduced by Hoek (1994) and developed by Marinos and Hoek (2000). Marinos et al. (2005) further discussed its applications and limitations.
The GSI system was extended to heterogeneous rock masses, such as flysch, by Marinos and Hoek (2001), and then modified by Marinos (2007), and Marinos et al., 2007, Marinos et al., 2011a with adjustments in values and additions of new rock mass types. Flysch formations are thus classified into 11 rock mass types (I–XI) according to the siltstone–sandstone proportion and their tectonic disturbance. Hence, a new GSI diagram for heterogeneous rock masses such as flysch has been presented, where a certain range of GSI values for every rock mass type is proposed (Fig. 3). It is highlighted again that the Hoek–Brown failure criterion and consequently the GSI value should be used when the rock mass behaves isotropically.
The case in the presence of better quality blocks along with the sheared mass may improve the “overall” rock mass strength, depending on their location and size. In the case where strong sandstone blocks are numerous and continuous and are with defined geometry, the rock mass properties can be evaluated by different approaches. Such an approach, the block in matrix approach (beamrocks), has effectively described by Wakabayashi and Medley (2004).
Basic inputs of the Hoek–Brown failure criterion, apart from the GSI value, are the uniaxial compressive strength (σci) and the material constant (mi) that is related to the frictional properties of the intact rock. Furthermore, in order to calculate the rock mass deformation modulus Erm, Hoek and Diederichs (2006) proposed a new equation, which includes the intact rock deformation modulus Ei, the GSI value and a disturbance factor due to the excavation method or a distressed character of rock mass D. Values of characteristic geotechnical parameters likely to prevail, for every flysch rock mass type (I–XI), are presented in Table 1. These values are resultant from the Roclab application (Rocscience Inc.). They are only indicative, since they cannot replace the detailed examination and the application of engineering judgement needed for each site-specific project separately.
The higher σci values are presented in sandstone flysch with a mean value of 45–50 MPa. In siltstone flysch, a mean σci value of approximately 15–20 MPa is promised. When the Ei is considered, a mean value of around 13 GPa is measured for sandstone flysch and 45 GPa for siltstone flysch (Marinos and Tsiampaos, 2010). Estimation of the mechanical parameters of a sheared siltstone or shale is a difficult task since the strength of the intact parts can hardly be measured in the laboratory (Fig. 4, Fig. 5). Representative strength values can, however, be assessed by back analysis (Tsatsanifos et al., 2000, Marinos et al., 2006b).
In addition, it is necessary to take into account the parameters of the “intact” rock properties σci, mi and Ei, and considerer the heterogeneous rock mass as a unit. Some quantitative estimates of heterogeneous intact rock properties via laboratory tests (Mihalis et al., 2010) have already been reported. In cases when laboratory tests are not feasible, a “specific weighted average” of the intact strength properties of the strong and weak layers was proposed by Marinos et al. (2011a).
The influence of groundwater upon the mechanical properties of the intact rock components, more particular on shales and siltstones that are susceptible to changes in moisture content in tunnelling is very important and has to be considered in the estimation of potential tunnelling problems.
Flysch, a typical impermeable formation, has the character of presenting alternations of strong brittleness with weak rocks. The latter strongly influences the development tendency of permeability due to the fracturing in the strong beds. Data collected in Northern Greece from 213 packer tests from 108 boreholes during site investigation for 8 tunnels in flysch environment showed the permeability values of about 4.5 × 10−7 m/s (Marinos et al., 2011b). The difference of different flysch types is very small, which can be explained with respect to the tectonic history of the flysch formation where a “homogenization” has achieved from the compression and folding process. The low values in the sandstone type are imposed by the barriers of the thin interlayers of siltstones, which may also intrude in major fractures of the sandstone beds. The decrease in relation to depth is progressive but with significant scatter (Marinos et al., 2011b). As a result of the low permeability, the water is not easily drained and it reduces the effective stresses and thus the shear strength of the rock mass. Many of these materials will disintegrate very quickly if they are allowed to dry out and not supported immediately.
Engineering geological behaviour during tunnelling
A further classification of flysch rock masses based on their geotechnical behaviour (deformation due to overstressing, overbreaks or wedge failure, “chimney” type failure, ravelling and their corresponding scale) is presented hereafter. Flysch, depending on its type, can present a variety of behaviours: being stable even under a noticeable overburden depth, exhibiting wedge sliding and wider chimney type failures, or showing serious deformation even under low to medium overburden. Its behaviour is basically controlled by its main geotechnical characteristics, considering of course the in situ stress and groundwater conditions. The study of the varying behaviours of various flysch types was based on the large set of data from the TIAS database.
After the identification of the failure mechanism, the suitable design parameters can be selected according to the principles of the failure mechanism. If the behaviour of the rock mass can be considered as isotropic and is governed by stress-induced failures, the user must focus on rock mass parameters. On the other hand, if the principal behaviour type is gravity-controlled failures (e.g. wedge sliding, chimney failures, ravelling ground), the user must focus on parameters related to discontinuities. If the rock mass is weak but also anisotropic (e.g. due to schistosity or well defined bedding planes), both the rock mass parameters and the persisting joint properties must be considered.
A reliable first estimate of potential problems of tunnel strain can be given by the ratio of the uniaxial compressive strength σcm of the rock mass to the in situ stress po (Hoek and Marinos, 2000). This is usually followed by a detailed numerical analysis of the tunnel’s response to sequential excavation and support stages. The strain estimation for the weak flysch rock mass type X of 4 different tunnel covers is shown in Fig. 6. It is evident that minor squeezing (category B) can be developed in the very poor flysch rock mass types X and XI from 50 m to 100 m tunnel cover, while severe to very severe squeezing (categories C and D) from 100 m to 200 m cover. Undisturbed rock mass types of sandstone or conglomerate (types I and III) do not exhibit significant deformations under 500 m.
More analytically, the strain estimation for one of the weakest flysch type for 4 different tunnel covers is shown in Fig. 7 (strain categories A–E according to Marinos and Hoek (2001)). An overstressed support shell due to squeezing is presented in Fig. 8, Fig. 9.
The presence of better quality blocks along the sheared mass may improve the stability of the surrounding rocks, depending on their location and size. A tunnel driven through this geomaterial requires continuous geological and geotechnical characterisation, as well as state of the art monitoring, to comprehend the complex interaction of internal block/matrix structure and their impact on the excavation and can only be conducted during tunnel construction. Such an effort was described in Button et al. (2004).
As far as the rheological characteristics of flysch formations are concerned, the creep potential of the sandstone formations is considered to be negligible. On the other hand, in the case of tunnel excavation in siltstone or shale formations, especially under high overburden, a time-dependent displacement or loads should be developed.
A detailed presentation of the range of geotechnical behaviour in tunnelling for each flysch rock mass type (I–IX) based on engineering geological characteristics is presented in Fig. 10. Generally, the behaviours of the flysch formations during tunnelling depend on 3 major parameters: (i) the structure, (ii) the intact strength of dominant rock type and (iii) the depth of the tunnel. The expected behaviour types (stable, wedge failure, chimney type failure, ravelling ground, shear failures, squeezing ground) can be illustrated in a tunnel behaviour chart (TBC) (Marinos, 2012). The main failure mechanism for every flysch rock mass type (I–XI) is projected in a TBC chart in Fig. 11.
Apart from the characterisation in Fig. 10, Fig. 11, the estimation of the tunnel behaviour and the philosophy of the support measures should be also performed on the basis of a detailed ground characterisation. This detailed characterisation cannot ignore the geological and/or in situ characteristics dictating or influencing the tunnel behaviour compared with a standardised classification (Marinos, 2012). This characterisation, named “Ground Characterization, Behaviour and Support for Tunnels” (Marinos, 2012) prompts user to evaluate the data in detail in order to assess the tunnel behaviour and adopt the appropriate support measures. An example of this characterisation in a tectonically disturbed flysch types is presented in Fig. 12.
The rock mass is often considered as an equivalent “mean isotropic geomaterial”, where rock mass properties are quantified through classification systems. This assumption is usually acceptable in cases of uniformly jointed, highly tectonised or disintegrated rock mass without persisting discontinuities of stable orientation controlling the rock mass behaviour. This is the case of the types VII–IX. In the case of bedded rock masses, at a scale of the tunnel section, the engineering geological behaviour during tunnel construction is significantly controlled by the characteristics of the stratification planes. This case may apply to flysch rock mass types IV–VI. A simulation of this anisotropic behaviour was analysed in Fortsakis et al. (2012).
Temporary support measures
The implementation of empirical tunnel design methods based on rock mass classification or simplified methods such as the convergence–confinement method should be of limited use in the design of tunnels in most of the flysch rock mass types. Such design cannot deal adequately with issues of face stability and the sequential excavation and installation of support. Therefore, the design of tunnels in weak flysch rock masses must involve the use of numerical methods. In some critical cases, like the simulation of the effectiveness of forepoling, tunnel advance and sequential support installation, three-dimensional numerical models should be used. However, in weak rock masses, the uses of sound engineering judgement and experiences from similar cases are valuable for the design and the construction of tunnel. The geotechnical properties of the material used for these analyses were calculated based on Hoek–Brown failure criterion. It should be highlighted here that in most of all cases the results of the model studies have been validated by the interpretation of convergence measurements and by the observation of the tunnel and installed support performance. Detailed principles and guidelines for selecting the immediate support measures are proposed based on the principal tunnel behaviour mode and the experiences from these 12 tunnels. In terms of permanent support concerned, different systems were presented in Fortsakis et al. (2004).
The tunnels under consideration are large in size with span of about 12 m. Apart from some cases of straightforward tunnelling in areas of good rock masses of flysch (types I–V), most of the studied tunnels were excavated under difficult geological conditions (types VII–XI). These tunnels have been excavated using top heading and bench method. Special measures were taken to stabilise the face like forepoling or/and installation of long grouted fibreglass dowels in the face. In addition, immediate shotcreting and leaving a core for buttressing have been used in different combinations for face stabilisation. After the stabilisation of the face, the application of the primary support system, consisting of shotcrete layers, rockbolts, steel sets or lattice girders embedded in the shotcrete in various combinations was necessary to ensure the stability of the tunnel. Elephant’s foot and micropiles in rare cases were used to assist the foundation of the top heading shell and to secure stability when benching. Temporary and permanent invert closure was implemented in order to face squeezing conditions. A typical support design for weak flysch rock masses, using top heading and bench method, is presented in Fig. 13 (Marinos et al., 2006a).
Under severe squeezing, the application of yielding systems was an alternative solution. The applied system was described in Schubert (1996) and Hoek et al. (2008). In the case of tectonically sheared siltstone rock masses under high cover (e.g. up to 250 m), where tunnel squeezing is a significant problem, the pillar stability in these twin tunnels requires careful evaluation.
The wide range of engineering geological behaviour leads to a corresponding range of temporary support measures. The temporary support in the specific tunnels discussed here varies from very light to very rigid or yielding. Temporary support measures concept and principles for every rock mass type are presented, based on the available tunnelling experiences, as shown in Fig. 14. It is not in the scope of this paper to provide analytical support measures. This work requires detailed design analysis of the tunnel support, adapted to the in situ conditions and particularities of each project. Here, the support proposals are reasonable considerations of both the rock mass behaviour and the critical failure mechanism, which are different for every flysch rock mass type. The necessity, the amount and the combination of the various elements of this typical section are results of numerical analysis and the optimization is a matter of reliable monitoring. The time of constructing temporary support is related with the support principle. A quick construction of a stiff support is usually implemented in case that there is a very small tolerance for displacements, whereas a yielding support that decreases the loads corresponds to a larger time interval.
The average excavation step for the top heading excavation of flysch rocks is presented in Fig. 15. The excavation step must be decided upon: (i) the anticipated size of wedges in the case of not tectonically stressed rock masses, (ii) the size of the wedges and the loosening prevention of the structure, in the case of disturbed rock masses without deformation problems, (iii) the prevention of structure loosening and (iv) decrease of deformation in association with the other appropriate measures in the case of weak rock masses where significant deformation is anticipated. For the cases (i)–(iii), the installation of spiles allows the increase of the excavation step. Excavation step is very difficult to exceed 1–1.5 m in very weak rock masses, while a mean value for the undisturbed rock masses could be 3 m.
The cost (Euros/linear metre of tunnel) of the temporary support system for the flysch formations from the experience of the Egnatia highway tunnels is projected in Fig. 16. This cost is presented in accordance with the “weight” of the support category.
The processing and evaluation of a great amount of geological and geotechnical information, obtained from the design and construction of 12 tunnels driven in flysch in Northern Greece, contributed to assessing the behaviours of the ground and the formulation in association with the correlations between ground and the formulation behaviours and the temporary support required.
Flysch formations are generally characterised by strong heterogeneity in the presence of low strength and tectonically disturbed structures, which may produce heavily sheared and chaotic masses. Flysch rock masses can be composed of sandstone and siltstone beds (undisturbed to folded) and inherently weak materials subjected to strong shearing where the original structure of the rock mass is no longer recognizable. The rock mass strength parameters needed for design can be sufficiently estimated by the Hoek–Brown failure criterion as long as the rock mass reacts isotropically to the underground excavation. Thus, a specialised GSI chart for the heterogeneous rock masses such as flysch can be used.
Flysch of various types can either be stable even under noticeable overburden and exhibit wedge sliding and chimney type failures, or cause serious deformation even under low to medium overburden. The rock mass behaviour in undisturbed to moderately undisturbed structures is highly anisotropic and controlled by the orientation and properties of discontinuities, mainly the bedding, in relation to the orientation of the tunnel. As a result, there is a possibility of wedge detachment and sliding along thin siltstone layers with low shear strength. The behaviour of the disturbed structures and even more of the heavily sheared rock mass types is generally isotropic, controlled by their low strength and low modulus of deformability. These masses may develop a significant deformation, even under low to medium overburden, while at greater depths squeezing prevails.
A wide range of temporary support can be applied in flysch rock masses, varying from very light to very rigid or yielding under severe squeezing conditions. Specific suggestions for the theory of temporary support in tunnel excavation through each flysch type are presented. These proposals take into account both the rock mass behaviour and the critical failure mechanism, which yet cannot replace the detailed analysis. They should be always back-analysed by engineering judgement and adjusted for each site-specific project.
Source: Tunnel behaviour and support associated with the weak rock masses of flysch