A slow and complex landslide under static and seismic action

Abstract

A dam construction project for the heightening of a dam is threatened by a large landslide located in the Spanish Southern Pyrenees. A marked syncline structure, folding a flysch formation overlying a marl substratum, is the main geological feature of the unstable valley slope. The landslide is characterized by two main and superimposed slip surfaces which follow the bedding layers. The motion observed is very slow and its acceleration can be related to rainfall regimes and changes in loading condition due to excavations. The paper provides an insight into the complexities associated with stability and kinematic evolution of a landslide under static and dynamic actions. Simplified procedures are proposed to predict the complex landslide behaviour. A comprehensive hydrogeological model is developed to relate rainfall history and water pressure which incorporates available data about the anisotropic permeability of flysch and the natural drainage obtained from pumping test and insitu measurements. The stability analysis of the slope focuses on the discussion of safety factors and their practical implication not only in terms of static limit equilibrium, but also on creeping velocity. The latter was evaluated under the simplified assumption of planar landslide and invoking the strain-rate effects on residual strength. A derived analytical expression provides a relationship between Safety Factor and creeping velocity. Under such hypothesis, the increment of the strength required to avoid predicting unrealistic landslide acceleration should be imposed for a lower range of shear velocity than the usual value imposed in the laboratory shearing tests. The seismic response of the landslide is also analysed. A Newmark’s approach and a stress-strain finite element (FE) analysis are compared, with special emphasis on the effect of the superimposed slip surfaces on the dynamic landslide response. To do that, the Newmark’s analysis has been extended to two overlaying sliding surfaces with independent strength properties and pore water pressure. It is concluded that assuming a single deeper sliding surfaces may underestimate the co-seismic permanent displacement. The comparison of two-block Newmark’s analysis with two-dimensional elastoplastic FE approach highlights the effect of geometry in the landslide motion.