Predictive maintenance of shield tunnels

• A comprehensive methodology for predictive maintenance of shield tunnel.
• FMEA approach to prioritize possible defects and facilitate the decision making.
• System-level lifing analysis for proactive maintenance of a tunnel system.
• Both risk prediction and damage accumulation models for system lifing analysis.
• Demonstrate with data for six failure modes collected from real-world tunnels.

Scientific maintenance methodologies are specially needed to enhance system reliability and safety, reduce maintenance manpower, spares, and repair costs, eliminate scheduled inspections, and maximize lead time for maintenance and parts procurement. This paper systematically presents a comprehensive methodology and framework for aggressive inspection and condition based predictive maintenance of a shield tunnel consisting of prefabricated lining rings. The framework consists of six components: maintenance purposes, data, modeling and simulation, documentation, managerial schedule, and inspection/maintenance behaviors. These components are interpreted in details as nine necessary parts in a predictive maintenance strategy. The failure mode and effect analysis approach is employed in developing a predictive maintenance strategy for a tunnel structure, with the purpose of prioritizing possible defects in the tunnel in order to facilitate the decision making on predictive maintenance. The system-level lifing analysis method is proposed for the proactive maintenance of the tunnel system. The method includes data preprocessing, risk model establishment, quantitative model validation, empirical lifing analysis, and system-level maintenance schedule. The empirical lifing analysis involves both risk prediction and damage accumulation models for service limit determination, system-level risk analysis, and system-level conditional risk for maintenance schedule. The proposed methodology is demonstrated with the inspection data collected for six typical defects observed in real-world shield tunnels.