Real-time ultra-deep mining geotechnical hazard prediction using statistical algorithms

This project aims to develop, and implement a code for real-time geotechnical hazard assessment and reporting for ultra-deep mining. This pilot project will be tested on a real mining site the Glencore’s Nickel Rim South Mine near Sudbury. This algorithm will represent a step-change in the capability to assess and manage geotechnical risk in mining, which will have particular value in the high-stress geotechnical operating conditions of ultra-deep mines. The problem of geotechnical hazard assessment is amenable to risk identification methods developed recently in the field of “predictive analytics”, one of the most research intensive areas in computational science. Preliminary tests of a predictive analytics system carried out for rockburst hazard assessment at a deep Sudbury nickel mine demonstrated remarkable success in hazard predictability relative to previously deployed statistical methods. A major aspect of this research proposal is detailed testing and refinement of this method for the geohazard problem, focusing on creation of a continuous learning system that will refine hazard assessments in response to mining history. If successful, this will represent a fundamentally new and powerful approach to geotechnical hazard assessment in ultra-deep mines.

Lorenzo Perozzi
Faculty Supervisor: 
Erwan Glaoguen
Partner University: