A temperature- and stress-controlled failure criterion for ice-filled permafrost rock joints

Abstract

Instability and failure of high mountain rock slopes have significantly increased since the 1990s coincident with climatic warming and are expected to rise further. Most of the observed failures in permafrost-affected rock walls are likely triggered by the mechanical destabilisation of warming bedrock permafrost including ice-filled joints. The failure of ice-filled rock joints has only been observed in a small number of experiments, often using concrete as a rock analogue. Here, we present a systematic study of the brittle shear failure of ice and rock–ice interfaces, simulating the accelerating phase of rock slope failure. For this, we performed 141 shearing experiments with rock–ice–rock sandwich' samples at constant strain rates (10−3 s−1) provoking ice fracturing, under normal stress conditions ranging from 100 to 800 kPa, representing 4–30 m of rock overburden, and at temperatures from −10 to −0.5 °C, typical for recent observed rock slope failures in alpine permafrost. To create close to natural but reproducible conditions, limestone sample surfaces were ground to international rock mechanical standard roughness. Acoustic emission (AE) was successfully applied to describe the fracturing behaviour, anticipating rock–ice failure as all failures are predated by an AE hit increase with peaks immediately prior to failure. We demonstrate that both the warming and unloading (i.e. reduced overburden) of ice-filled rock joints lead to a significant drop in shear resistance. With a temperature increase from −10 to −0.5 °C, the shear stress at failure reduces by 64 %–78 % for normal stresses of 100–400 kPa. At a given temperature, the shear resistance of rock–ice interfaces decreases with decreasing normal stress. This can lead to a self-enforced rock slope failure propagation: as soon as a first slab has detached, further slabs become unstable through progressive thermal propagation and possibly even faster by unloading. Here, we introduce a new Mohr–Coulomb failure criterion for ice-filled rock joints that is valid for joint surfaces, which we assume similar for all rock types, and which applies to temperatures from −8 to −0.5 °C and normal stresses from 100 to 400 kPa. It contains temperature-dependent friction and cohesion, which decrease by 12 % °C−1 and 10 % °C−1 respectively due to warming and it applies to temperature and stress conditions of more than 90 % of the recently documented accelerating failure phases in permafrost rock walls.