A 3D Implementation of an Elastoviscoplastic Model and Application to Creep-Driven Slow-Moving Landslides

To analyse problems involving creep-driven slow-moving landslides interacting with structures and infrastructure and gain useful information for risk management, numerical modelling is the most comprehensive tool. When resorting to numerical simulations, appropriate constitutive modelling is key to capture time-dependent behaviour of geomaterials. However, a drawback of advanced constitutive models accounting for time-dependency is the high number of parameters involved, which are commonly calibrated based on site-scale monitoring data, without accounting for the characteristic creep behaviour of the material. This paper describes the implementation of an existing elastoviscoplastic model in a three-dimensional finite-difference software. The model, called SHELVIP (Stress Hardening Elastic Viscous Plastic), was derived from the classical theory of elastoplasticity by adding a viscoplastic component based on Perzyna’s overstress theory. After illustrating the model and the implementation procedure, the steps to tune the constitutive parameters, especially the viscoplastic ones, are described. First, the viscoplastic parameters are determined at the laboratory scale through an analytical calibration process based on laboratory creep tests. Then, tuning of some parameters at the site scale is performed, while other parameters are kept unchanged from their laboratory-scale value being representative of the intrinsic creep behaviour. The final set of parameters can be used to run predictive analyses for problems of engineering relevance. The analytical calibration is exemplified with reference to triaxial creep tests on Palombini clays. Then, the implemented SHELVIP model is validated by comparing numerical, analytical and experimental creep strains at different deviatoric stress levels. A three-dimensional numerical model of the interaction between a deep-seated landslide and an existing motorway tunnel is developed. The capability of the constitutive framework to catch the time-dependent landslide movement is demonstrated by comparing monitoring data with the corresponding numerical results. The model is then used to run predictive analyses aimed at investigating the slow-moving landslide-infrastructure evolution.