Webinar: Modelling Rock Strength With Machine Learning and Probabilistic Methods

Webinar Details

A practical webinar exploring how machine learning and probabilistic analysis can support rock strength modelling workflows in IP.

Rock strength is an essential part of any geomechanical model. However, rock strength prediction is often hampered by limited data or no data at all - leading to substantial uncertainty in the geomechanical models. While access to extensive rock strength datasets can result in more robust models, uncertainty can never be eliminated entirely.

This webinar will discuss two contrasting cases: one where hundreds of UCS scratch test data is available, and another where little to no data exists. Rock strength models will be built for both scenarios, and the resulting uncertainties will be examined and compared.

• Building UCS prediction models using machine learning
• Comparing Domain Transfer Analysis, Neural Networks, and Multiple Linear Regression
• Optimizing model performance through intelligent feature selection
• Understanding the limits of model transferability across formations
• Developing rock strength models with minimal or no core data
• Applying deterministic and probabilistic uncertainty analysis
• Quantifying the impact of rock strength uncertainty on safe mud weight windows
• Reducing uncertainty through drilling observations and data acquisition

Speaker 

Frans Mulders

A long-standing member of the Geoactive team, Frans is an experienced Senior Engineer with a demonstrated history of working in operations, applied and fundamental research, and consultancy in Europe and the Asia Pacific Region. With a PhD in Technical Geoscience from Delft University of Technology (Netherlands) and MSc in Geology / Engineering Geology from RWTH Aachen University (Germany), there isn't much Frans doesn't know about the application of Geomechanics in a subsurface environment.