A step toward better understanding rock bolt technology

A step toward better understanding rock bolt technology

This thesis aims to advance the performance, reliability, and design of rock bolting systems by developing an integrated framework that combines experimental investigation, numerical simulation, and machine learning alongside the development of a novel generation of fibreglass rock bolts. Motivated by the need for safer, more durable, and cost-effective reinforcement technologies in mining, civil, and geotechnical engineering, this study presents a multidisciplinary investigation that spans systematic literature review, laboratory experimentation, numerical modelling, machine learning approaches, and manufacturing process to develop comprehensive solutions across multiple domains of rock bolting technology. The systematic review mapped the evolution of research trends, collaboration networks, and key thematic areas within the field of rock bolting. It identified significant knowledge gaps concerning the influence of grout mechanical properties, rock bolt installation deviations, and the limited application of predictive modelling techniques such as machine learning methods. Experimental pull-out tests were conducted to evaluate axial load performance under varying grout conditions, and the results informed the development of machine-learning models that predict rock bolt performance with promising accuracy up to 98.9%. These models were further enhanced using ensemble learning techniques and optimised with metaheuristic algorithms, achieving an accuracy of 98.8%, which highlighted the importance of grout strength, type, and water-to-grout ratios as dominant variables. The study also explored the use of machine learning in estimating grout compressive strength, supported by an extensive experimental program on five different cementitious grout types. The influence of rock bolt eccentricity and inclination on load transfer was investigated through pull-out tests and numerical models, revealing that minor deviations can lead to significant reductions in performance. Additionally, innovative fibreglass rock bolts were developed using a braided truss manufacturing method, and shear testing demonstrated marked improvements, with a 26% increase in shear capacity and energy absorption compared to standard fibreglass rock bolts. The research findings highlighted the importance of integrated experimental, numerical, and machine learning approaches in advancing rock bolt technologies. New opportunities were proposed for optimising material properties, expanding predictive modelling capabilities, improving installation practices, and validating innovative designs through full-scale field trials.

File reproduced in accordance with the copyright policy of the publisher/author.