The Role Of Progressive Brittle Fracture In The 1931 Landslide At Dogface Rock, Katoomba

Zack Tuckey

The 1931 Dogface Rock landslide in Katoomba NSW was a complex, progressive cliff collapse with a failure volume in the order of 100,000 m3 that was triggered by the extraction of remnant coal pillars from the Katoomba Colliery, about 200 m below the top of the escarpment. Although underground coal mining is generally accepted as a cause of the rockslide, previous studies have not explicitly investigated the role of progressive brittle fracture in the collapse. This paper presents an integrated study which incorporates remotely piloted aircraft photogrammetry with a discrete element method numerical investigation of the landslide, and thereby explores the role of progressive brittle fracture, and re- examines the failure mechanism and runout motion of this multi-stage landslide.

Remotely piloted aircraft photography is used to build a georeferenced 3D model of the site with Structure-from-Motion photogrammetry software. A digital geotechnical mapping workflow is demonstrated to investigate the morphology of the landslide scar, extract statistics on discontinuity orientation, persistence, and spacing, and undertake trace mapping of newer brittle fractures that interacted with pre-existing high persistence joints as the landslide rupture surface developed. A series of discrete element method numerical laboratory tests are used to calibrate bonded block contact properties that reproduce laboratory scale intact rock index parameters including UCS and tensile strength. Upscaled rock block contact parameters are then applied to a cliff-scale model that investigates the progressive development of rock mass damage induced by mining. Following extraction of the remnant pillars, rock mass damage develops mostly by extensile strains that produce tension cracks. Brittle fractures propagate upwards from the mine level and eventually initiate toppling of massive sandstone slabs defined by high persistence pre-existing subvertical joints. The investigation illustrates how the integration of photogrammetry with discrete element numerical methods can be used to characterise progressive brittle failure and runout of large rock slope failures.