ABSTRACT
Modeling of natural ground conditions will always require approximations due to their spatial variability. Often limited datasets describing material behavior are available resulting in the estimation of material properties. In many cases only deterministic estimates of material strength are reported or used in geotechnical analysis, despite the natural variation and uncertainty of material strength in the area under investigation. Failure to take into account the random nature and variation of material mechanical strengths can result in misleading stability calculations. This paper shows the difference in slope stability results that are calculated when soil properties are modelled as deterministic values, interpolated from known sample sites, described by a range of values (probabilistic analysis) and varied spatially and stochastically.
