Abstract
Predicting the longitudinal response of existing tunnels subjected to shield undercrossing remains challenging because of uncertainties in excavation-face deformation and excavation-induced changes in the supporting soil stiffness. This study develops an analytical framework that accounts for both non-uniform excavation-face contraction and local subgrade stiffness degradation. The framework consists of two stages. First, tunnelling-induced greenfield soil settlement is calculated using a quasi-stochastic medium formulation, in which non-uniform excavation-face contraction is represented by the superposition of uniform radial contraction and vertical translation. A dimensionless parameter is introduced to characterize the contraction pattern and treated as an uncertain input, allowing an envelope of possible tunnel responses to be obtained. Second, the existing tunnel is modelled as a Timoshenko beam resting on a Pasternak two-parameter foundation, with excavation-induced ground settlement represented by an equivalent distributed load and local passive subgrade stiffness degradation described by a Gaussian-type empirical relationship. The governing equation is solved using a finite difference method with a matrix formulation accommodating different end boundary conditions. The method is verified against centrifuge test results and field measurements from the Shenzhen Metro Line 3 under-crossed by Line 9. In both cases, the predicted settlement envelopes encompass the measured maximum settlements of approximately 4.5 mm. Neglecting local stiffness degradation underestimates the maximum settlement in the centrifuge case by approximately 13%. Parametric analyses further show that the maximum normalized settlement decreases by approximately 55% as the normalized tunnel separation increases from 1 to 5, whereas it increases approximately proportionally with the soil-loss ratio over the investigated range. The influence of non-uniform contraction becomes more pronounced when the tunnel separation is smaller and the soil-loss ratio is higher. The proposed framework provides an efficient analytical approach for evaluating existing tunnel responses and their uncertainty under shield undercrossing.