Mechanical behaviour of Clay-Shales (Argille Scagliose) and implications on the design of tunnels

In the framework of Geotechnical Engineering and Rock Mechanics many cases are known (described in either the classical or the modern tunnelling bibliography) where, even during construction, large deformations and high stresses in the lining are observed. This is the result of natural phenomena identified with swelling and/or squeezing conditions. This thesis is devoted to the study of tunnelling in difficult conditions, with particular attention to the development of large time-dependent deformations. These deformations may develop either during the construction stage, causing instabilities of the tunnel face and unsafe working conditions, or remain hidden during the “short-term”, thus leading to complex problems when the tunnel is put into service. Under these circumstances the construction costs may rise due to the delays in excavation time, the stabilising and heavy support measures to install. Though the physic-chemical reactions which are the basis of these phenomena have been studied for nearly one century, during the last thirty years the research work concentrated on the rigorous identification, quantification and prediction of the models of behaviour associated with them. In particular, during the last few years, two Research Programmes on these topics (“Tunnelling in difficult conditions” and “Mechanised excavation of tunnels”, co-ordinated by Professor Giovanni Barla) were carried out with the financial support of the Italian Ministry for University and Research. Into this contest, a Ph. D. thesis [Barla M., 1999] was dedicated to the determination of the typical stress-paths around a tunnel and the application to laboratory testing by means of a newly developed triaxial cell (Soft Rocks Triaxial Apparatus, SRTA). The samples of a stiff clay came from the Caneva-Stevenà quarry, near Pordenone (Italy), where swelling induced deformations and instabilities caused failure of the 30 cm thick unreinforced concrete lining of the adits. The decision to further investigate these phenomena came from the growing world-wide interest on the subject and from the availability of a new relevant case study. Several cubic samples of tectonized clay shales (Chaotic Complex) were, in fact, obtained at the face of Raticosa tunnel and Osteria access adit, along the new high speed railway line Bologna-Florence. The present study deals with the models of behaviour which were proposed in recent years, with the aim to identify the most significant factors involved in the selection and design of stabilizing measures of the tunnel face and supports to be installed along the heading. The testing programme performed on the clay shales had the task of identifying similarities and/or differences with respect to the previous research. Moreover, the characterisation studies performed allowed for the determination of significant influence of expandable minerals, complex structure and low mechanical properties of these geomaterials. The testing programme included triaxial tests performed on natural material in closely controlled conditions; oedometer type tests were performed on reconstituted samples. The SRTA was modified for the purpose of testing the new material. Other characterization tests (Atterberg limits, X-ray diffraction analyses, oedometer tests on natural material, etc.) were performed in other laboratories (Enel-Hydro, Milan and Seriate). An extended and critical bibliographic study was carried out with the intent of describing the principal available methods for modelling and predicting the swelling behaviour and their consequences. Though several methods are presently available in literature, they are often very specific and effective for particular case studies. Moreover, the models of swelling behaviour are often embedded in relatively simple elastic and elasto-plastic constitutive laws, which do not allow one to take into account time-dependent deformations. Further investigation in the field of advanced models was carried out, with the intent of determining the significant factors influencing the real mechanical behaviour. The second part of this study was devoted to the numerical modelling at the sample scale and at the tunnel scale. Numerical analyses were performed by the Finite Difference Method and an axi-symmetric coupled model reproducing the sample behaviour at laboratory scale. The stress-paths of the triaxial tests reproduced the behaviour of a point located on the tunnel sidewalls during construction with initial isotropic state of stress. Three elasto-plastic laws were assumed for the ground: Drucker-Prager, modified Cam Clay and Nova-Lagioia model [Nova & Lagioia, 2000]. The stress-strain curves obtained from the laboratory tests could be represented in a satisfactory way, but the corresponding excess pore pressure at low stress level (positive) or at incipient failure (negative) could not be predicted reliably. Since any attempt of reproducing the tectonized clay shales behaviour by means of either simple or more complex elasto-plastic models has been shown to be not effective, an alternate way for the purpose of modelling was represented by time-dependent models of visco-elastic plastic type. Considering that the clay shales exhibit at low stress level a significant time-dependent response, the visco-elastic plastic Burgers’ model with Mohr-Coulomb yield criterion (CVISC) as available with the FDM code Flac [Itasca, 1999] was used. The parameters describing the time-dependent response of the specimens (as evidenced during the laboratory tests) were determined by means of closed-from solutions (Burgers’ visco-elastic model) and numerical analyses with the CVISC constitutive equation. In both cases a satisfactory description of the mechanical behaviour could be achieved. Numerical analyses were performed with the intent of reproducing the deformational response of the Raticosa tunnel, for which monitoring data were available (radial and longitudinal displacements). Numerical analyses were performed by the FDM and an axi-symmetric coupled model reproducing the full excavation sequence of an equivalent circular tunnel. The geotechnical parameters entering the CVISC model for the clay shales at the tunnel scale needed to be assessed in terms of the parameters obtained at laboratory scale. In fact, the parameters based on laboratory testing were not likely to reproduce the tunnel behaviour as observed during excavation. This was the case for deformability, strength, and time dependent parameters, which were evaluated on the basis of experience and in situ observations. The tunnel response in terms of plastic zone extension, radial and longitudinal displacements versus time could be reproduced satisfactorily. Based on of the present work, the following conclusions can be drawn: (a) the time-dependent behaviour of the clay shales is a significant factor to be taken under close consideration for the assessment of the tunnel response to excavation (the elasto-plastic constitutive laws are not capable of reproducing the full range of behaviour of the tunnel); (b) laboratory testing is necessary in order to determine the relevant features of the mechanical behaviour of the material, however the parameters determined from laboratory tests cannot be directly used for appropriate prediction of tunnel behaviour; (c) monitoring is essential for the assessment of the tunnel response (stability of the face and of the core ahead of the face), including the effectiveness of the stabilization measures and of the primary lining, and the time of its installation