Numerical and probabilistic analysis of reinforced soil structures

Ioannis Zevgolis, Purdue University

Abstract

Reinforced soil structures have become in many occasions an attractive alternative to traditional reinforced concrete retaining structures. An application that has been employed in recent years is their use as bridge abutments, where they simultaneously function as earth retaining and load bearing structures. This research addresses two important issues that are currently omitted from conventional consideration of this type of structures. First, a plane strain finite element analysis is performed to investigate the performance of the system abutment - reinforced soil structure - foundation soil with respect to deformations. An elastic perfectly plastic Mohr-Coulomb model and a plasticity hardening hyperbolic model are used to capture the behavior of granular and compressible material, respectively. Using effective stress parameters, both drained and undrained conditions are simulated, the latter one accounting for full development of excess pore pressures. Analyses are performed for a variety of loading conditions, geometric characteristics, and foundation profiles, taking into account stage construction. For the conditions examined in the present study, the results indicate the absence of differential settlements that are often observed when deep foundations are part of the system. In addition, it is shown that the loads from the bridge superstructure play only a minor role in the anticipated immediate and consolidation settlements. The second issue addressed by this research, is the development of a probabilistic model in order to assess the reliability of reinforced soil structures. Geotechnical uncertainty is explicitly considered by modeling shear strength properties as random variables following beta distributions. The model is based on Monte Carlo simulations and accounts for the dependency between the involved failure mechanisms. External stability is approached as a system in series, while internal stability is modeled in two stages: a series configuration addresses the reliability per reinforcement layer, and then an r-out-of-m configuration is used to model internal stability as a system. A framework for consideration of redundancy and propagation of failure is also formulated, based on transitional probabilities and Markov stochastic processes. As an illustration, a case example of a reinforced soil structure used as direct bridge abutment is analyzed in the context of the model developed.

Degree

Ph.D.

Advisors

Bourdeau, Purdue University.

Subject Area

Geotechnology|Civil engineering

COinS