A Time-Dependent Viscoelastic Cohesive Zone Model and Inversion Method for Analyzing Interface Damage of Embedded Tram Track

Abstract

The cohesive failure between the asphalt pavement and the rail wrapping material around the tram track is the one diseases of the new embedded tram track structure. A time-dependent viscoelastic cohesive zone model (CZM) was employed to characterize interface behavior between asphalt pavement and rail wrapping materials. By integrating Maxwell rheological elements into a bilinear CZM framework, the model captures time-dependent traction–separation behavior. Key features include distinct stiffness evolution during elastic deformation and relaxation-driven traction variations under different loading rates (10–300 mm/min). An Elman neural network surrogate model was developed to inversely identify five critical interface parameters from experimental load–displacement curves, achieving high accuracy (RMSE: 0.0143–0.2384, R² > 0.9). Validation via interface pull-off test demonstrated strong agreement between simulated and experimental results, confirming the model's efficacy in predicting viscoelastic interface degradation. This framework provides a robust tool for analyzing time-sensitive cohesive failures in urban rail infrastructure.