Deterministic and random FEM analysis of full-scale unreinforced and reinforced embankments

This paper presents the results of numerical simulations of one physical full-scale unreinforced embankment and two reinforced embankments taken to failure using a strip footing. The embankments were 3.4 m in height and were constructed with the same slope using the same sand backfill. One reinforced embankment was constructed with a relatively weak and extensible geogrid material and the other with a relatively stronger and stiffer geogrid product. Deterministic finite element method (FEM) and random finite element method (RFEM) analyses were carried out and the results compared to observed performance. Analyses were performed assuming vertical spatial variability in soil strength at a scale of fluctuation matching the compaction lift thickness of the physical tests and treating the soil as a random (homogeneous) material. Both deterministic and probabilistic analyses showed good agreement with the measured bearing capacity of the footing in each full-scale test. However, the assumption of vertically anisotropic spatial variability in soil strength was shown to achieve a better match with the observed failure mechanism in each of the embankments.