3D Numerical Analyses of Column-Supported Embankments: Failure Heights, Failure Modes, and Deformations

Design of column-supported embankments (CSE) requires the evaluation of global stability using the conventional limit equilibrium method (LEM). Yet, for CSEs using unreinforced concrete columns and load transferring geogrids, the failure mechanisms and corresponding soil-structure interactions are not well understood. There is increasing evidence pointing to large bending moments in columns and failure of columns in flexure, as opposed to a failure by shear as assumed in limit equilibrium analyses. In response to these design uncertainties, the failure height, failure mode, and deformations of eight column-supported embankment scenarios were investigated using three-dimensional (3D) numerical analyses. For the same embankment scenarios at failure height, factors of safety (FS) were then calculated using the two-dimensional (2D) LEM for investigating its applicability in evaluating global stability of CSEs. The 3D numerical analyses examined CSE stability for the limiting conditions at undrained end-of-construction and after long-term dissipation of excess pore water pressures. The numerical model included representations of flexural tensile failure in the concrete columns and tensile failure in the geosynthetic reinforcement. Scenarios consisted of a base case with typical concrete column design, five single-parameter variations using base case conditions, and two multiparameter variations using base case conditions. The undrained condition was the most critical, and two failure modes were found: (1) multisurface shearing in the embankment coupled with bending failure of columns and near-circular shear failure in the clay, and (2) multisurface shearing in the embankment coupled with bending failure of columns and shearing in the upper portion of the soft foundation clay. Both failure modes were accompanied by a rupture of the geosynthetic when included in the load transfer platform. Soil-column interactions were complex, and many columns failed in bending at lower embankment heights than those that produced collapse. The factors of safety calculated using the LEM were overstated. This is because the LEM assumes failure by shear, which has limited applicability for examining the complex mechanisms by which CSEs fail. The practical implication is that the LEM should not be used for evaluating global stability of this system type and, by extension, other system types in which soil-structure interactions result in failures controlled by mechanisms other than shear.