Computational issues in anthropogenic land subsidence prediction

Anthropogenic land subsidence due to fluid withdrawal is usually predicted with the aid of Finite Elements (FE) in both the coupled and the uncoupled formulations. Nowadays FE models to simulate land motion have become quite a well-established technique, requiring advanced tools to improve the overall quality of the analysis. Present computational issues include: a) the implementation of cost-effective and accurate iterative solvers relying on unsymmetric preconditioned conjugate gradients properly stabilized to make the use of coupled models attractive in large complex geological structures; b) the development of special interface elements yielding a more accurate stress/strain analysis in faulted porous media; c) the use of stochastic models to address the mechanical heterogeneity of the depleted formations. Representative numerical examples are shown and discussed to illustrate the importance of the above computational issues.