Abstract:
Recent tailings dam failures around the world have highlighted the real risk posed by undrained slope failures. Undrained failures are fundamentally different to drained failures in the sense that different mechanisms are involved (i.e. a slope may be stable against drained failure but unstable against undrained failure). Popular methods to assess the stability of slopes against undrained failure involve the use of limit equilibrium analyses with both drained and undrained strengths assigned in the same analysis. A potential shortcoming of these methods is that no consideration is given to strain compatibility. In this study, a limit equilibrium based method where strain compatibility is maintained on the failure surface was developed. The method, referred to as the strain mobilisation method, considers a Mohr circle of stress at failure to determine the shear strength mobilised on the failure plane for use in the stability analysis as a function of the deviator stress imposed on the triaxial test result. The mobilisation of stress on a failure plane with strain was determined based on the stress-strain relationships observed during triaxial tests. A Factor of Safety (FoS) was used to express the stability of the slope as a function of the mobilised strain and the calculated FoS results obtained using the proposed method were compared to calculated FoS results using traditional methods. This was done for three tailings materials (gold, iron and platinum) for three specific hypothetical slopes. As an additional check, the proposed method was tested on Nerlerk sand, a well-known sand showing strain-softening behaviour during undrained shearing. It was found that, in general, as mobilised strain is increased, the FoS calculated using the proposed method converges to that of traditional methods so that there was no significant difference in calculated FoS between the current methods and the proposed method that does consider strain compatibility. This indicates that the proposed method provides FoS values comparable to those calculated using currently accepted methods where the failure surface passes predominantly through a single material type. For such a case, there does not appear to be a need to consider limit equilibrium methods where strain compatibility is maintained. The proposed method provides an indication of the amount of strain that may be expected to mobilise to provide the FoS. Given that this amount of strain is not excessive, the current methods which do not consider strain compatibility perform satisfactorily and can continue to be used
