Capturing and characterising pre-failure strain on failing slopes

Abstract

Effective management of slope hazards requires an understanding of the likely triggers, geometry, failure dynamics, mechanism and timing; of these the last two remain most problematic. Reducing the epistemic uncertainty of these elements is crucial, particularly for landslides that are not easily mitigated. The ‘inverse-velocity method’ utilises the linearity in inverse-strain-rate change through time in brittle materials to forecast the timing of final slope collapse. A significant body of published deformation data is available, yet to date there has been no attempt to collate a catalogue of landslide deformations from a large number of sites to examine emergent behaviour; notably variations in and controls on movement prior to failure. This thesis collates thirty-one examples of tertiary creep and related attributes from a broad literature search of over 6,000 peer-reviewed journals. Results show that tertiary creep operates over durations ranging from ~37 minutes to 3,171 days. Patterns of acceleration corroborated with published parameterisations of brittle failure; namely Voight’s (1989) model. Most examples (86%) were best-fit with hyperbolic curves, described by an α coefficient within the 1.7 and 2.2 range; indicative of deformation driven by crack growth. No significant relationships between slope and creep characteristics were found within the database of examples, however the lack of standard reporting of slope failures, particularly between industry documents and academic papers, limits the analysis. The database validates the ‘inverse-velocity method’ as a robust forecasting technique. Iterative a priori analysis of data has shown that slopes deforming in a brittle manner are more likely to predict slope collapse ‘too soon’ as a false positive prediction. Analysis has also shown that tertiary creep is typically delimited (87% of examples) within the first 25% of the total creep duration. Recommendations towards monitoring specifically highlight the need for instruments to deliver spatial accuracies to ~10mm, surface based capture and continuous measurement. Developing processing procedures for point cloud data derived from a permanent terrestrial laser scanning system is recommended as the best approach to small-scale deformation monitoring.