Abstract
Current practice for evaluating the vulnerability of upstream-raised tailings dams involves the computation of factors of safety by the limit equilibrium method assuming drained and undrained shear strength and, in the case of brittle tailings, also assuming softening to residual strength.
In this latter case, a FoS = 1.0 may be accepted, in the rationale that if such FoS is attained, the dam would be stable even in the event of a full liquefaction of the tailings body. This method has drawbacks: i) it is deterministic, and its result depends on the choice of the liquefied strength, dragging uncertainty into the process and outcome; ii) it provides no information about which actions could lead to such massive liquefaction; and iii) it does not address the deformations that the dam might undergo in the event of a limited liquefaction.
To circumvent these drawbacks, the evaluation is complemented with trigger analyses, which can be broadly grouped in two types: i) simulating credible modes of failure by applying external actions that are relevant for a particular dam, e.g. a rise in the phreatic level in the dam´s body; and ii) external actions that are known to be deleterious are applied to the dam, regardless of their plausibility, and these actions are increased in value until the dam fails.
In this paper, recent experiences with trigger analyses of upstream-raised tailings dams are commented and some discussions on the uncertainty in the interpretation of the results are provided.
