Predicting brittle failure in rocks using machine learning

Applications are invited for a 3-year position in a Research Fellowship as PhD Candidate in Geosciences to be based at the Njord Centre, a cross-disciplinary research unit at the interface between Physics and Geosciences at the University of Oslo.

Image may contain: Atmosphere, Light, World, Astronomical object, Atmospheric phenomenon.

Machine learning prediction of brittle failure will involve data processing and numerical modelling.

When brittle materials are loaded, numerous microfractures may nucleate, grow, and coalesce before major catastrophic rupture occurs. This mechanical behavior controls the deformation of various materials such as cements, metals, ice, and rocks in the Earth’s crust where earthquakes occur. However, the origin and evolution of these microfractures, also called damage, are not understood and could be used to monitor and predict the occurrence of major earthquakes. Using machine learning and state-of-the-art rock physics experiments performed at the European Synchrotron Radiation Facility (ESRF, Grenoble, France), the candidate will participate to deformation experiments of rocks approaching failure (4D X-ray microtomography, acoustic emission monitoring). The participant will analyze these data using several computer science approaches, including supervised and unsupervised machine learning techniques. The work will also include numerical mechanical modelling of fracture network development using the Discrete Element Method. The main question will be to identify how the precursors to failure in rocks and other brittle materials localize on the future fracture plane when approaching failure and, to what extent water may control this process.

At the Njord Centre, the candidate will be supervised by François Renard, professor of Geosciences, and Dr. Jess McBeck, researcher in geophysics. The candidate will perform his/her main activities in the framework of CompSci and will collaborate with another PhD student performing experiments in the European Research Council (ERC) project "Break-Through Rocks".


  • MSc in Physics or Earth Sciences, preferably in rock physics with data processing experience.
  • Candidates with documented experience in X-ray imaging and/or ultrasonic monitoring techniques will be prioritized.
  • Skills in scientific programming (Python, Matlab), simulations, and experience from machine learning will be considered as important assets.
  • Fluent oral and written communication skills in English.

Personal skills

  • Teamwork skills, as well as the ability to work independently
  • Enthusiasm, personal drive, and ability to take initiative
  • Attention to detail
  • Flexibility and ability to meet deadlines


Professor Francois Renard

Researcher Bennoit Cordonnier

Call 2b: Project start winter 2022/2023

This project is in call 2b, starting winter 2022/2023. 

Published June 23, 2022 10:50 AM - Last modified June 23, 2022 10:50 AM