SME and MGS Technical Presentation and Dinner

While individual point load tests (PLT) are relatively quick and easy to conduct, full PLT campaigns and borehole drilling can be time-consuming and expensive. Machine learning techniques can be used on existing PLT and core logging data to make predictions of Is50 where point load tests haven’t been conducted, providing a quick and cheap method for obtaining large quantities of intact rock strength information. Rock mass strength can be subsequently estimated from this and used to populate a geospatial block model of rock mass strengths throughout the excavation. The PLT measurements and machine learning predictions allow us to better understand rock behavior and strength variability, crucial for optimizing the design parameters of future excavations.

Chris Thielsen M.S. is a geomechanics engineer with the Itasca Consulting Group, Inc. After graduating with a bachelor’s and master’s degree in geoengineering from the University of Minnesota, he started as a consultant with Itasca in 2021, focusing on the application of numerical modeling and machine learning to solving problems in geomechanics. His current work focuses on the creation of surrogate models with artificial neural networks and training random forest models on core logging data that are used to predict rock strength at open-pit and underground mines worldwide.

Host or Publisher SME Southern Minnesota Subsection