AI-assisted reliability-based design framework for tunnel concrete linings in weak rocks

Area of Science:

• Geotechnical Engineering
• Civil Engineering
• Tunneling

Background:

• Reliability-based design (RBD) for tunnel concrete linings in weak rock is complex due to parameter uncertainty and ground-support interaction.
• Existing methods often rely on deterministic safety factors, which may not adequately address inherent risks.

Purpose of the Study:

• To develop an integrated framework for the reliability-based design of tunnel linings in weak rocks.
• To provide risk-consistent design objectives beyond traditional safety factor criteria.
• To assess the performance of conventional and fiber-reinforced reactive powder concrete (FRPC) linings.

Main Methods:

• Implicit ground-support equilibrium analysis using the Convergence-Confinement Method (CCM).
• Probabilistic evaluation of safety factors, reliability index, and failure probability via Monte Carlo sampling.
• Development of an AI-driven surrogate model for rapid parametric analysis.
• Benchmarking against the First-Order Reliability Method (FORM).

Main Results:

• The integrated framework successfully links lining thickness and compressive strength to target reliability levels.
• Increased lining thickness and material quality (especially FRPC) significantly reduce failure probability.
• The AI surrogate model accurately predicts probabilistic trends with high consistency.
• FRPC offers substantial reliability improvements, particularly at intermediate thicknesses.

Conclusions:

• The developed CCM-reliability-AI framework offers a transparent and efficient approach to tunnel lining design in weak rocks.
• This methodology facilitates risk-consistent and economical design choices, enhancing tunnel resilience.
• The study highlights the benefits of advanced materials like FRPC for improving tunnel safety and performance.