Discrete Fracture Network Optimization for Tunnel Digital Twin Management

Abstract

The internationalization of the research aims to advance infrastructure management, with a focus on tunnels, through the integration of advanced digital modeling technologies and Bayesian optimization. Tunnels play a crucial role in transportation networks and societal well-being, yet the efficient management of these structures faces challenges due to aggressive environments and the complex interaction among support elements, rock masses, and service demands. In this context, characterizing rock discontinuities and optimizing digital models to predict structural behavior are essential. The digital twin (DT) approach offers a virtual representation connected to reality, mirroring the behavior of its physical counterparts. Although the application of this technology in civil infrastructure is still in its infancy, tunnels represent an ideal asset for technological advancements. The connection between performance loss assessments, lifecycle considerations, and decision-making process optimization through DTs represents significant advancements in civil infrastructure management. The project aims to incorporate information about rock discontinuities into the DT model of the tunnel under analysis, using Discrete Fracture Network (DFN) optimized by advanced mathematical models. Additionally, the project seeks to connect the responses to structural pathologies, recorded by inspections, to the DFN model defined for the tunnel, thereby optimizing deterioration prediction and lifecycle management. To achieve these objectives, advanced Bayesian optimization and stochastic DFN modeling techniques will be employed, with the collaboration of Prof. Cacciari, a specialist in rock mechanics. The project development will occur in stages over a six-month period, with activities including data organization, Bayesian optimization programming, model integration, and scientific report elaboration. Collaboration with Polytechnique Montréal will provide access to advanced resources and expertise in civil and geotechnical engineering, strengthening international connections and contributing to project internationalization. The project will benefit from the university's robust investment in high-performance software from ITASCA, seamlessly integrated with Rhinoceros, the Digital Twin's development environment. Additionally, high-resolution 3D printers will be used for potential testing at reduced scale, and access to clusters for high-memory processing analysis, essential due to the computational demands of simulating behavior in DFN models. The results obtained will be disseminated through scientific publications and knowledge transfer to research groups in Brazil, thus expanding the impact and visibility of the work conducted.