On 4 August 2014, a breach occurred in the perimeter embankment at the Mount Polley Tailings Storage Facility in British Columbia, Canada, causing a spill of mining waste into the environment. The Government of British Columbia retained an Independent Review Panel (2015) to determine the cause of failure.

The breach was sudden and without observable precursors. During the collapse, the mass of soil underwent a rotational-translational movement involving large horizontal displacements in a foundation unit ~10m below original ground level. The slippage at the base took place in a thin (≤2m) deposit designated as the “Upper Glaciolacustrine Unit”, or the Upper GLU. The IRP determined that undrained strengths controlled this unit’s mechanical behaviour during failure. Furthermore, the clay’s strain-weakening properties made it susceptible to progressive failure. The IRP found that the breach occurred when the peak undrained shear strength of this material was exceeded. These findings were supported by two-dimensional analyses.

A detailed three-dimensional static analysis demonstrated that, due to large amounts of shearing resistance developed along the sides of the slide, the entire Upper GLU area involved in the failure would have to fully weaken in order to bring the soil mass to a limiting equilibrium. Such a result posed two additional questions, one related to pre-failure deformation levels and another pertaining to the failure modes in the Upper GLU.

Laboratory tests indicate that shear strains ≥60% would be required for the unit to fully weaken;​ in a 2m deposit, this may mean lateral deformations ≥1.2m prior to collapse. From the brittle nature of this failure we know that no such deformations had taken place. Additionally, deformation analyses have shown that a portion of Upper GLU in the failure zone, about ⅓ by area, would have remained overconsolidated during collapse and thus much stronger.

To reconcile the apparent incongruity of conclusions suggested by static and deformation analyses, it has been hypothesized that (a) the Upper GLU strained non-linearly, weakening considerably prior to collapse, but without significant shear displacements;​ and (b) some other material was weaker at failure than originally thought. The rockfill material in the shell area was a suspect due to poor compaction during placement. A proposition was put forward that the rockfill’s deformation modulus was significantly lower than that of other materials involved in this failure. In the absence of substantial deformations prior to collapse, this material was thought to have only partially mobilized its shear strength.

The failure at Mount Polley was investigated using three-dimensional deformation analysis. The mechanical behaviours of soils involved in the failure were captured through customized constitutive models that were developed on the basis of, and calibrated against, laboratory testing results and published data. The embankment construction sequence was simulated in nine loading stages.

The simulation results indicate that the progressive failure at Mount Polley started as early as 2011, advancing as the embankment construction proceeded, but remaining contained until the summer of 2014. By the fall of 2013, in addition to the contained failure, three specific material conditions developed in the foundation materials that brought the structure to the verge of instability. These are (a) the substantial depletion of reserve shear strengths in the materials surrounding the plastic yield zones;​ (b) the emergence of a large area close to the precipice of weakening;​ and (c) the extension of some brittle soil units in the failure zone.

In the final construction stage, the addition of 2.5-4m of embankment materials in the shell, crest and beach areas triggered collapse under undrained conditions. The collapse unfolded in two distinct phases. In phase one, the failure processes were largely contained to a thin shear band in the Upper GLU where ongoing strain-weakening processes resulted in a decrease of shear resistance and an accumulation of shear displacements. In phase two, the shear zone propagated into other soils, and multiple local failures developed in the upstream region of the slide. In this phase, a sustained drop of mobilized shear resistance levels was observed at the base and in the upstream regions of the slide. In the shell zone, the shear strength of the rockfill was not fully mobilized even in the advanced stages of collapse.