Analytical Graphical Approach for Predicting Ground Conditions in TBM-based Tunneling Construction

Abstract

The present master's thesis addresses the use of Artificial Intelligence (AI) and Machine Learning (ML) algorithms to predict geology based on Tunnel Boring Machine (TBM) data. The use of mechanized tunneling has become frequent over the last decade, and their performance is critical for project management and safety. Numerical simulation methods have become prevalent in predicting TBM performance metrics, and the use of AI/ML techniques for prescient applications using TBM-generated data has become ubiquitous. The current research aims to propose an exploratory look into the correlation between specific TBM parameters and ground conditions. The methodology seeks to classify rings based on three main ground classes: rock, soil, and mixed, through the observation of clear patterns, found to be representative of these ground classes, which are demonstrated. A techno-economic assessment of the current use of AI/ML tools for geology prediction in TBM-based tunneling construction, is also presented, analyzing both the potential and shortcomings of the technology. For the purpose of the study, the Porto Metro project (Portugal) is introduced, used as a case study for the proposed methodology. As the mining and drilling market is projected to almost double from 2020-2030, and with the increasing use of TBMs, improving ground condition prediction is paramount to the advancement of tunneling automation efforts. The present thesis aims to further develop the field and open dialogue on the use and effectiveness of using purely AI/ML modelling methods for this application.