Professor Kyelgyenbai on caving: AI-ready curricula for remote mine engineers
Reviewed by Tom Sullivan

First reported on Australian Centre for Geomechanics – News
30 Second Briefing
Autonomous deep-level mass caving at operations such as Oyu Tolgoi is driving a shift from tactile, face-level decision-making to remote system control, where engineers manage autonomous LHD fleets, ventilation-on-demand and microseismic arrays from virtual control rooms. Professor Khavalbolot Kyelgyenbai argues mining curricula must integrate AI, ML and control theory so students can interpret real-time data from thousands of sensors, apply stress–extraction relations such as Es = f(Ve, ΔK)·e^(σ1/σ3), and run neural-network-based hazard forecasting. Simulator-centric labs mimicking remote operation centres, with multi-screen cave digital twins and injected seismic or mudrush events, are positioned as core training tools at the Mongolian University of Science and Technology.
Technical Brief
- Microseismic energy release is modelled as Es = f(Ve, ΔK) · e^(σ1/σ3) to couple extraction rate and stress.
- Neural networks are trained on historical rockburst and cave-collapse datasets to detect non-linear precursors to failure.
- Genetic algorithms and reinforcement learning agents are used to compute optimal draw sequences across hundreds of drawpoints.
- Simulation labs inject events such as localised seismic shocks, mud inflows and automation communication blackouts into cave digital twins.
- Students must diagnose anomalies, reroute autonomous fleets and reschedule draw to maintain geotechnical and operational safety envelopes.
- Research methods centre on AI/ML model development using multivariate sensor streams, historical hazard records and synthetic cave simulations.
- Practical application is direct: decision-support for draw control, airblast/mudrush risk flagging and real-time cave stability assessment.
Our Take
Oyu Tolgoi’s underground performance has been strong enough to lift Rio Tinto’s Q1 copper output, according to our April 2026 coverage, so any caving research from the Mongolian University of Science and Technology that improves draw control or stability directly supports a high‑throughput, high‑stress operation rather than a greenfield concept.
Recent disruptions to Oyu Tolgoi concentrate haulage by the Radical Reform Movement show how social licence issues in Mongolia can quickly translate into export risk; stronger, locally led research on caving safety and ground control gives operators additional credibility when negotiating with communities and regulators around underground expansion.
Within our 1265 Mining stories, only a subset of base‑metals pieces are tagged to both Research and Safety, so work like Professor Kyelgyenbai’s on caving in Mongolia sits in a relatively specialised niche that practitioners can use as a regional reference point rather than relying solely on case studies from Chile or Canada.
Prepared by collating external sources, AI-assisted tools, and Geomechanics.io’s proprietary mining database, then reviewed for technical accuracy & edited by our geotechnical team.
Related Articles
Related Industries & Products
Mining
Geotechnical software solutions for mining operations including CMRR analysis, hydrogeological testing, and data management.
Tunnelling
Specialised solutions for tunnelling projects including grout mix design, hydrogeological analysis, and quality control.
CMRR-io
Streamline coal mine roof stability assessments with our cloud-based CMRR software featuring automated calculations, multi-scenario analysis, and collaborative workflows.
HYDROGEO-io
Comprehensive hydrogeological testing platform for managing, analysing, and reporting on packer tests, lugeon values, and hydraulic conductivity assessments.
GEODB-io
Centralised geotechnical data management solution for storing, accessing, and analysing all your site investigation and material testing data.