TL;DR
This paper introduces an IoT-based monitoring system and a Bayesian-genetic algorithm to analyze the mechanical behavior of large-diameter underwater shield tunnels during construction. By inverting measured internal forces, the study provides a more accurate estimation of external earth and water pressures compared to traditional methods, offering crucial insights for ensuring tunnel safety.
What Problem Does This Paper Address?
Constructing large-diameter underwater shield tunnels is a complex process where the tunnel segments experience significant and dynamic loads from the surrounding soil and water. Traditional design methods often simplify these conditions, particularly during the transitional phase from segment assembly to full load bearing. This can lead to an inaccurate assessment of the actual stress state, potentially compromising the tunnel's long-term safety and reliability. The paper addresses the challenge of accurately monitoring and analyzing these complex mechanical behaviors and external loads during the critical construction phase, especially when the tunnel is still exposed to fluctuating conditions before the surrounding soil fully stabilizes.
Methods
The study employs a multi-pronged approach:
- IoT-based Monitoring System: An on-site monitoring system utilizing embedded Fiber Bragg grating (FBG) sensors was established. These sensors are strategically placed within the tunnel segments to measure parameters like earth pressure, pore water pressure, concrete stress, steel stress, and bolt stress. This system allows for real-time data acquisition during the construction process.
- Bayesian-Genetic Algorithm for Load Inversion: A novel inversion analysis model was developed using a Bayesian-genetic algorithm. This algorithm is designed to estimate the external earth and water pressures acting on the tunnel segment. It works by:
- Establishing a relationship between the segment's internal forces (measured by sensors) and external loads.
- Utilizing a genetic algorithm to optimize parameters related to the segment's mechanical response.
- Employing Bayesian optimization to fine-tune the superparameters of the genetic algorithm, enhancing computational efficiency and accuracy.
- Elementary Beam Theory: Standard elementary beam theory is used to calculate the bending moment and axial force of the segment ring based on the strain measurements from the concrete strain gauges.
Datasets
The study utilizes data collected from an on-site monitoring system installed on a specific test ring (R600) of the Qinwang tunnel project in Zhejiang Province, China. This dataset includes measurements of:
- Earth pressure
- Pore water pressure
- Axial force
- Bending moment
This data was collected during the critical stages of segment assembly and removal from the shield tail. The paper does not introduce a new benchmark dataset but rather uses field-collected data from a real-world engineering project.
Key Insights and Ideas
- Dynamic Load Fluctuations: The construction process, particularly segment assembly and removal from the shield tail, induces significant fluctuations in external loads and internal forces. These fluctuations are more pronounced than in a stable operational state.
- Peak Load Magnification: The peak external load observed during the fluctuation stage is approximately twice that of the stable stage. Similarly, internal forces like axial force and bending moment can be significantly higher (up to eight and five times, respectively) during the fluctuation phase compared to the stable phase.
- Importance of Construction Phase Analysis: The study highlights that focusing solely on operational loads is insufficient. The mechanical behavior during construction, especially the transitional periods, can have a substantial impact on the tunnel's long-term performance and requires careful consideration.
- Inversion Analysis Superiority: The proposed Bayesian-genetic algorithm inversion method demonstrates a significant improvement in accuracy for estimating earth and water pressures compared to traditional full coverage pressure theory. This suggests that field-measured internal forces are a more reliable basis for load estimation.
- Limitations of 2D Theory: The discrepancy observed between calculated and measured axial forces suggests that a purely 2D theoretical approach may not fully capture the complex three-dimensional mechanical effects of the segment, especially those influenced by longitudinal jacking thrust and inter-segment friction.
Key Results
- The peak external load during the fluctuation stage was found to be twice that of the stable stage.
- The peak axial force during the fluctuation stage was eight times higher than the stable value.
- The peak bending moment during the fluctuation stage was five times higher than the stable value.
- The inversion analysis method showed a 51.16% reduction in calculation error for earth pressure compared to the full coverage pressure theory.
- The inversion analysis value for earth pressure was 76.21% higher than the measured value, while the full coverage pressure theory was 153.04% higher.
- For water pressure, the inversion analysis value was 17.86% higher than the measured value, compared to 5.58% higher for the full coverage pressure theory.
Limitations and Future Work
- Limited Measured Data: The authors acknowledge that the accuracy of the load inversion model is currently limited due to the relatively small amount of measured data.
- Discrepancy in Axial Force: A significant disparity exists between the calculated and measured axial forces, attributed to the limitations of the 2D theory in capturing 3D mechanical effects.
- Future Work: The researchers plan to explore data augmentation methods to enhance the model's accuracy. Further research is needed to address the limitations of 2D theory and improve the prediction of axial forces.
Summary updated