Open Collections

Abstract

Various failure modes exist for tailings storage facilities (TSFs). Some credible failure modes are unique to one or a few sites, though many are common to nearly all facilities. With the adoption of the Global Industry Standard on Tailings Management (GISTM), the definition of credible failure modes and the requirement to minimize failure risks to “as low as reasonably practicable” have often increased the need for a thorough evaluation of failure risks. The effort to evaluate at least some failure modes may be substantially reduced by developing fragility models. This approach reduces the likelihood of a failure mode to a set of key inputs or characterizations. A multivariate function can be developed based on historical analyses and/or with supplemental analyses to evaluate the sensitivity of results to various inputs. This function establishes a relationship between a hazard intensity measure, such as peak ground acceleration or wind speed, and a performance parameter, such as damage or loss. This relationship is known as a “fragility function.” This paper develops a set of fragility functions for a TSF failure mode involving landslide-induced impulse waves, which could lead to the overtopping of a TSF embankment. After identifying and sampling hazard intensity measures (e.g., the landslide volume) and variables (e.g., the density of liquified tails) that affect the landslide-induced waves, various scenarios are analyzed to determine the resulting overtopping damage and develop the corresponding fragility functions. Similar failure modes are commonly evaluated in water reservoirs, where impulse waves have led to severe overtopping events (e.g., the Vajont Dam, which overtopped in 1963 in northern Italy). However, the analysis becomes more complicated in TSFs where waves would propagate over an exposed beach. The impulse wave failure mode is used as an example to show the ease and usefulness of implementing the fragility method to rapidly assess failure modes that could otherwise require significant effort. Lastly, the practicality of the methodology is showcased through its application to a real-world copper TSF, demonstrating its effectiveness in estimating failure probabilities associated with specific hazard scenarios.