Probabilistic analyses of geosynthetic-reinforced pile-supported embankments using design methods and 3D finite element models considering soil variability

Abstract

Geosynthetic-Reinforced Pile-Supported Embankments (GRPSE) are effective composite structures to support highway infrastructures on weak soils. Numerous design methods have been developed in practice to facilitate the use of this technology. However, it is well known that as the design methods adopt different theoretical assumptions, the performance indexes given by the design methods vary significantly. Furthermore, the effect of soil variability on the design outcomes given by the design methods is unknown. These uncertainties present great challenges to design engineers to select the proper design method and consider soil variability. To fill this knowledge gap, we conducted comprehensive probabilistic analyses using typical design methods (BS8006 and CUR226) and 3D finite element models (unit-cell and full-scale) considering soil variability. A well-established case study in the literature was used as a benchmark. Algorithms for the design methods and the 3D finite element models were developed, calibrated and tested in both deterministic and probabilistic scenarios. A detailed probabilistic comparison between the design methods and the 3D finite element models was also carried out. Results show that 1) soil variability affects the performance indexes of GRPSE, including stress reduction ratio, stress concentration ratio, differential settlement, and tensile force of geosynthetics; 2) model uncertainties of design methods can be as high as 46%, due to the assumptions and simplifications made to formulate the solutions; 3) the probabilistic 3D full-scale finite element method is the most robust approach to consider soil variability.