SBD-K-medoids-based long-term settlement analysis of shield tunnel

Long-term settlement is essential to operational tunnel safety. In this study, a systematic method for analysing shield tunnel settlement during the operation period is proposed. First, a new time-series clustering algorithm, namely shape-based distance (SBD)-K-medoids, is proposed to mitigate the limitation of low efficiency. The clustering precision and efficiency of the new algorithm were validated through hypothesis testing. Subsequently, the long-term settlement analysis framework is illustrated in detail, consisting of four parts: data pre-processing, section division and settlement characterisation, similarity clustering, and attribution analysis. A new index, called the incremental tunnel settlement state (ITSS), is proposed to depict a shield tunnel's long-term longitudinal settlement during the operation period. Finally, a case study of the Shanghai Metro Line 10 (ML10) was conducted to validate the method. The results indicate that the uplink of ML10 is divided into four clusters, which correspond exactly to four situations a shield tunnel section undergoes during operation. Through an attribution analysis, the formation conditions of each cluster are summarised. Corresponding settlement patterns are concluded and engineering countermeasures are proposed for rehabilition. In general, this case exemplifies the rationality and engineering practice value of the study's approach, and the systematic method offers a new approach for analysing the long-term settlement of shield tunnels.