Effect of a fractured zone on drift behaviour in the argillaceous Tournemire site (France).

G. Armand, A. Rejeb Rockmeck 2002 - VI th Regional Rock Mechanics Symposium, SELCUK University, Konya, Turquie - 9-11/10/2002.

The French Institute of Radioprotection and Nuclear Safety (IRSN) has selected the Tournemire site to increase their knowledge on nuclear waste storage in argillaceous media. This site is a good study case for modelling since numerous field and laboratory experiments have been conducted there, and especially because two similar drifts have been excavated, one in an unfractured medium and the other in a fractured one. This study is a preliminary step for a full-scale 3D model of the site, in which different sections have been modelled to study the drift convergence in the fractured zone. Parametric studies were conducted, with three different mechanical behaviours, and the results are here compared to the in situ measurements. This study is the preliminary step for a complete 3D modelling of the Tournemire site to better understand how to model the fault and the fractured zone. Two different zones, a fractured one and an unfractured one, separated by a principal fault, were considered in 2D sections . The fractured zone is considered as an homogeneous zone with reduced mechanical properties in terms of elastic and plastic properties. Different simple behaviours such as isotropic linear elasticity, orthotropic linear elasticity and elastoplasticity were tested and the calculated drift convergences were compared to the in situ measured ones. The elastic calculations show that the convergence is slightly sensitive to Poisson's ratio changes, but very sensitive to the Young's modulus changes. Laboratory tests show an orthotropic behaviour. If the measured orthotropic parameters are used in the model, the convergence in the unfractured case is underestimated. For the orthotropic behaviour the Young's modulus in the fractured zone must be divided by four to obtain the same convergence. The in situ stress measurements show some dispersion of the stresses, that why the choice of initial condition is important. If the extreme stresses values are taken into account, the K0 coefficient varies between 0.5 and 1.5 and the average convergence between 1 mm and 7 mm. The fault was model by elastic joint elements. The difference between the results calculated with and without elastic joint elements is significant when the fault crosses the gallery or is close to the gallery. If the fault is not close to the gallery, no matter how the joint is modelled, the convergence is not modified. The thickness of the joint element (from 0.5 m to 1 m) does not affect this result. A decrease in normal stiffness increases tunnel convergence more than a decrease in shear stiffness. The convergence and extension of the plastic zone is not affected by a change in the plasticity limit of the far field. The plastic zone stretches close to the tunnel due to the material properties in the neighbourhood of the tunnel. The plastic zones in each sections, except C3 where the drift passes through two different materials, are symmetrical around the tunnel. The perfect Drucker Prager model, used with the compressive strength measured in laboratory test, gives no plastic zone. Other elastoplastic laws, especially with a non-associated flow rule, will have to be tested. The fractured zone must be regarded as a damage zone with reduced mechanical properties. Further study will have to be conducted on the elastoplastic behaviour of the rock masses and the fault. In particular, an orthotropic elastoplastic criteria with a non-associated flow rule seems to be better suited to modelling the behaviour of the fractured zone.