Slope Stability Analysis Techniques Incorporating Uncertainty in Critical Parameters

The use of "probability" methods in slope stability analysis has been practised by geotechnical engineers for some time. This paper questions whether the term probability has been properly used since in many cases the term is used in a classical statistical manner only to describe frequency distributions of data. This paper proposes a definition of the term probability which includes the engineers' subjective state of confidence for all critical parameters. Such parameters may include the slope stability model, undetected structural features, strength, groundwater pressure distribution, and geological framework. This process does not make a "better" analysis but it does force the geotechnical engineer to state his opinion in such a manner that others can understand and if necessary, criticize the assumptions. Further, this technique forces the question, "What impact would failure of a slope have on the production schedule or costs of a mine?" The level of risk taken in adopting a particular slope design strategy in a mine is not a question of what Factor of Safety should be selected but, rather, what trade-offs result from steeper slopes and the associated costs of failure. This paper also presents several simple probabilistic techniques which, though widely used in other fields have been all but ignored in geotechnical slope stability analyses. These include the use of stratified sampling to optimize Monte Carlo uncertainty approaches, and the use of simple functional approximations rather than complex stability models in probabilistic analysis. The author presents a case example of this technique and discusses the results in comparison with more traditional methods.