Browse by author
Lookup NU author(s): Dr Gaetano EliaORCiD
The behaviour of earth dams subjected to seismic action has traditionally been studied adopting semi-empirical approaches or simplified numerical methods such as the limit equilibrium analysis or the equivalent visco-elastic scheme. More sophisticated numerical tools, based on the finite element (FE) method, have also been developed in the academic environment to simulate the dynamic behaviour of large earth embankments (e.g.: Griffiths & Prevost, 1988). In the last decade the FE approach, based on fully coupled effective stress formulation for the solid-fluid interaction, is becoming more and more popular in the engineering environment as a tool for the design of earth dams, allowing to predict the serviceability and stability of such structures (Zienkiewicz et al., 1999). Any reliable FE analysis of the dynamic response of an earth dams should be based on an accurate description of the stress-strain behaviour of the material during the seismic action. With this respect, a number of experimental results have indicated that the essential features of the mechanical behaviour of soils subjected to cyclic loading are state dependency, early irreversibly, non linearity, decrease of nominal stiffness and related hysteretic energy dissipation. Only some of these features can be simulated adopting simple constitutive assumptions in the framework of elasto-plasticity (e.g.. Elastic-Mohr-Coulomb model, Elastic-isotropic hardening Modified Cam-Clay model). A primary limitation of these single surface models in the context of dynamic analyses is related to their inability to reproduce the observed hysteretic dissipation. This is caused by the unrealistically large extent of their yield surfaces, which leads to limited accumulation of plastic strain during cycles. As a consequence, the use of such simple models in FE dynamic analyses requires the superposition of a, often large, fictitious viscous damping. This last ingredient of the analysis is difficult to quantify a priori and, at the same time, can have a crucial influence of the results (Woodward & Griffiths, 1996). More advanced constitutive models can be used in FE dynamic analyses of large dams in order to properly predict all the main features of the soil behaviour mentioned above. Despite the more complex mathematical formulation, these models are able to simulate the early development of irreversible strains and the permanent pore water pressures built up, thus resulting in a realistic prediction of the hysteretic dissipation and, as such, limiting the amount of the required additional viscous damping to a low value, no longer significantly influencing the results. The aim of this work is to emphasise the merits of the fully coupled non-linear approach, based on the use of a sophisticated FE code and of an advanced constitutive model, in the prediction of the behaviour of an ideal homogeneous earth dam subjected to a real earthquake. In particular, the same problem is first analysed assuming for the soil a two surfaces mixed isotropic-kinematic constitutive model and then re-analysed adopting the relatively simple Modified Cam-Clay model. The results of the two analyses are compared in terms of acceleration profile, time history and Fourier spectra, together with displacements at the crest, excess pore water pressures and stress paths. In the final part of the paper, a discussion of the results and some brief conclusions are proposed.
Author(s): Elia G, Amorosi A, Chan AHC
Publication type: Conference Proceedings (inc. Abstract)
Publication status: Published
Conference Name: 11th International Conference of IACMAG
Year of Conference: 2005
Date deposited: 20/01/2010
