GeoMod

Internships @GeoMod in 2021 in Lausanne: probabilistic methods or app development a distancia, residencia en América del Sur

Internship: Probabilistic methods in computational mechanics with application to geomechanical failure analysis

Duration: 4 to 6 month, fulltime, in 2021

Motivation: Probabilistic computational mechanics made considerable progress in the past 20 years, in many aspects, random fields, reliability and bayesian approaches in particular. While probabilistic considerations are often underlying safety concepts in existing civil engineering norms, they do not enter systematically into the safety evaluation or the optimisation processes explicitly. The time for a stronger integration of probabilistic approaches into safety assessment procedures and computational mechanics in general seems appropriate.

Project: The proposed project will lay at the intersection of soil mechanics, computational failure analysis, and probabilistic methods. Based on the work of earlier trainees, who looked into uncertainty propagation, sensitivity analysis and reliability analysis, the present project will focus on a bayesian approach. The trainee will be asked to investigate the potential, for a given geotechnical problem, resulting from the coupling of an existing probabilistic toolbox, UQLab developed at ETHZ, with an existing failure analysis software, ZSOIL developed at Zace Services Ltd in Lausanne, both distributed worldwide. More specifically the trainee will be asked to: assimilate the work already done by former trainees on uncertainty propagation with UQLab and ZSOIL, reproduce results taken from the literature relative to a specific geomechanical problem., formulate and implement a bayesian approach.

Documentation: ZSOIL and UQLab

Prerequisites: MS student in computational mechanics or equivalent. Understanding of failure analysis, background in computational mechanics, strong interest in probabilistic methods.

Interested? Please contact Prof. S. Commend for further information.
Lausanne, 12/2020