A new approach to assess the stability of rock slopes and identify impending failure conditions

Anticipating the failure of unstable slopes is a topic of major concern in the field of landslide risk management. Arbitrary displacement or velocity thresholds are typically used to setup alarms representative of different risk levels. This is a difficult process, since experience showed that failure can occur over a wide range of values of these two parameters. Other approaches, such as the well-known inverse velocity method, aim at forecasting the time of failure; however predictions can be characterized by a significant margin of error. It follows that determining when the failure of a slope is impending is still a great issue. It is therefore important to review monitoring data from past case studies and seek for recurrent correlations between specific variables in order to identify common slope behaviors in proximity of the instant of failure. The analysis of a database of rock slope failures at several coal mines is presented. For each event values of cumulative displacement and velocity measured 48 h before failure, 24 h before failure, 3 h before failure and at failure instant were provided. A straightforward correlation was found between the average acceleration during the final 3 h and the average acceleration during the final 24 h before failure, regardless of the order of magnitude of the deformation. Comparable results were obtained considering data relative to 3/48 and 24/48 h before failure. Consequently, impending failure conditions could be determined with suitable notice for future events of slope instability at the studied mine sites based on the relative intensity of the acceleration. Further development of the methodology requires that the presence of similar correlations in other databases of past slope failures is verified. The nature of the correlation may be influenced by factors such as geology, failure mechanism and mechanical properties of the rock mass.