Smart segment-enabled IoT monitoring of shield tunnel longitudinal joints: Inferring hidden structural states from measurable segment rotations

Joints are vulnerable components in shield tunnels. As far as longitudinal joints are concerned, critical states such as opening angle and concrete damage remain hidden during operation. Measuring these states directly necessitates placing sensors at joints, but these sensors may malfunction under harsh conditions and even damage waterproofing systems. This research considers the issue of inferring hidden joint states from segment-embedded sensors located at protected positions rather than joints. This approach is built on physics-based transfer functions that relate measurable segment rotations to hidden joint states. Four steps constitute the framework. First, full-scale bending tests provide the basis for characterizing the nonlinear moment-rotation behavior of longitudinal joints, revealing distinct stages from elastic response to failure. These results provide physical ground truth for numerical calibration. Second, a segment-level finite element model is calibrated against the test results to establish a validated numerical model, including material parameters, contact settings, and boundary conditions. Third, ring-level parametric simulations inherit these calibrated parameters and vary surface surcharge to establish transfer functions from rotations at the embedded sensor locations to joint opening angle and concrete damage. Fourth, a reliability analysis framework incorporates load uncertainty to quantify probabilities of different safety states. The aforementioned protected sensors are realized through smart segments with embedded wireless tilt sensors. Integrated with wireless gateway and cloud platform, these Internet of Things (IoT)-enabled smart segments serve as core monitoring elements and enable continuous data acquisition from casting through operation. One-year field monitoring data are exploited to assess the framework. Inferred joint opening angle stabilizes at approximately −0.13°, corresponding to an equivalent surface surcharge of 73.3 kPa; under the assumed load distribution, the probabilities of healthy, warning, and critical states are 88 %, 8 %, and 4 %, respectively. This framework shifts tunnel monitoring from passive deformation observation to active structural state inference, providing a practical smart solution for preventive maintenance.