Analytical Modeling of Time-Dependent Railway Subgrade Settlement Driven by Stress Release During Shield Tunnelling

Shield tunnelling beneath existing railway subgrades is still a problem in geotechnical engineering, largely because rail infrastructure responds poorly to even small vertical movements. Track systems tolerate very little distortion before operational limits are reached, and relatively minor differential settlement can translate into service restrictions, intensive monitoring, or unplanned reinforcement during construction. With the increase in cities density and heavier use of underground space, tunnelling beneath active rail corridors is now common and this shift has exposed the limitations of prediction methods that focus only on the final settlement profile, without addressing how deformation accumulates as excavation advances.

Although settlement induced by tunnelling has been studied for decades, its temporal development is still not straightforward to predict. Empirical approaches remain widely used, in part because they are easy to implement and grounded in accumulated field experience. However, these methods depend on simplified descriptions of settlement troughs and offer little transparency regarding the underlying mechanics. Their reliability also deteriorates when ground conditions are modified through reinforcement, since key parameters lose their original physical meaning. Numerical simulations provide a more explicit representation of construction sequences and soil–structure interaction, but this comes at the cost of extensive calibration, long runtimes, and sensitivity to modelling choices that are often difficult to justify during preliminary design. Data-driven techniques have attracted attention more recently, yet their reliance on dense monitoring records and limited interpretability restricts their usefulness before construction begins.

Indeed, analytical methods continue to play an important supporting role because   they provide a level of transparency that is often absent from other approaches by expressing settlement directly in terms of mechanical actions and material response. Solutions derived from the Mindlin framework have been especially influential, as they allow forces associated with tunnelling to be treated as distributed loads acting within an elastic ground mass. Extensions of this idea have accounted for face pressure, shield friction, grouting effects, and layered stratigraphy, yielding predictions that compare favourably with observations under controlled conditions. Still, most existing formulations treat settlement as an instantaneous outcome of applied loads. The excavation process, however, unfolds progressively, and the ground responds continuously as stresses are redistributed over time which are not easy to resolve in analytical settlement prediction.

A further difficulty arises from how ground loss is treated. Conventional analytical and semi-empirical models require assumptions regarding soil convergence or volumetric loss ratios, parameters that become uncertain once reinforcement modifies stiffness and stress redistribution. These assumptions hinder accurate description of how settlement accumulates during excavation and consolidation phases. A recent research paper published in Tunnelling and Underground Space Technology and conducted by Professor Yao Shan, Dr. Guankai Wang, Mr. Weifan Lin, Professor. Shunhua Zhou from the Shanghai Key Laboratory of Rail Infrastructure Durability and System Safety, at Tongji University in collaboration with Professor Frank Rackwitz from the Technical University of Berlin, the researchers developed a time-dependent analytical method for predicting railway subgrade settlement during shield tunnelling that integrates construction actions. They introduced a stress-release representation of ground loss that replaces assumed convergence models with an equivalent load framework and the method combines multiple excavation-related effects within a single Mindlin-based formulation while remaining computationally efficient.

The research team formulated an analytical framework that represents shield tunnelling as a sequence of mechanically distinct actions applied to a reinforced, layered soil mass. The authors first translated complex reinforcement schemes into equivalent elastic parameters by drawing on composite material theory, which allowed reinforced strata to be represented without sacrificing directional stiffness effects relevant to vertical deformation. The investigators then simplified the stratigraphy into an equivalent single-layer system so that established elastic solutions remained applicable while preserving the influence of stiffness contrasts above and below the tunnel axis. The authors examined settlement contributions by explicitly associating each construction action with a corresponding stress representation. They also demonstrated how face thrust, shield–soil friction, and synchronous grouting pressure could be expressed as spatially distributed loads whose influence migrated with tunnel advance through coordinate transformation. By embedding tunnel velocity directly into these transformations, the team ensured that settlement at a fixed monitoring point evolved continuously as excavation progressed, rather than appearing as a static superposition.

An important contribution emerged from the treatment of ground loss because instead of prescribing volumetric loss or convergence geometry, the investigators modeled excavation-induced loosening as a gradual release of in situ stress around the tunnel periphery. The team examined how this released stress could be recast as an equivalent load acting on the surrounding soil once the shield tail passed a given section and by linking the stress release rate to tunnel position, the authors aligned settlement development with excavation progress in a manner consistent with observed behaviour. The researchers observed that this formulation naturally reproduced distinct settlement phases reported in monitoring data, including early uplift or minor deformation during cutterhead approach, rapid settlement during shield passage, moderated response during grouting, and continued deformation during post-excavation consolidation. When applied to both a published benchmark case and a coastal railway undercrossing project, the analytical predictions followed measured time histories closely, especially in capturing prolonged settlement after excavation ceased. The study demonstrated that settlement patterns near the tunnel axis were predicted conservatively, while lateral distributions matched observed trough shapes with increasing accuracy away from the centreline.

In conclusion, Professor Yao Shan and colleagues provided a pathway for analytical prediction that remains valid even when reinforcement alters stiffness and load transfer by treating excavation-induced loosening as progressive stress release.  This perspective strengthens the physical basis of settlement analysis and reduces reliance on empirical calibration tied to specific soil conditions. Beyond its immediate application to railway undercrossing, the framework clarifies how time-dependent settlement emerges from the interaction between excavation actions and soil response. The explicit inclusion of tunnel advance allows engineers to anticipate not only peak settlement but also its rate of development, information that is critical for operational decision-making such as speed restrictions or monitoring thresholds. The method’s efficiency makes it suitable for early-stage evaluation, where multiple alignment or reinforcement scenarios must be assessed rapidly. Moreover, the new approach of Professor Yao Shan and co-workers provides a template for extending analytical solutions to other tunnelling contexts where staged construction and stress redistribution dominate ground response. While the formulation remains grounded in elastic theory, it suggests a bridge between short-term excavation effects and longer-term ground behaviour. With the increase in demand transparent, interpretable prediction tools in infrastructure projects the study work reinforces the continuing relevance of analytically grounded models within modern geotechnical practice.