Treating ballast and subgrade soil as an integrated unit for sampling and loading has proven to be an effective method for investigating the interaction between ballast and subgrade soil. Given that direct testing of specimens containing large ballast is constrained by the capabilities of standard laboratory equipment, adopting a model material of smaller size is recommended. Parallel gradation method is widely used for this purpose. This study performed an evaluation of parallel gradation method based on the response of ballast penetration into subgrade soil. Discrete element models were developed to simulate the penetration of crushed ballast, featuring three different parallel gradations, into subgrade soil. On this basis, dynamic triaxial simulations were conducted on these models. By comparing the macroscopic and mesoscopic mechanical characteristics at different scaling ratio, the applicability of the parallel gradation method for assessing ballast penetration into subgrade soil was evaluated. At the macroscopic scale, the scaling ratio of crushed ballast significantly influences the axial, volumetric, and lateral deformations observed during penetration into subgrade soil. Specifically, a smaller average grain size of ballast correlates with reduced deformations in these specimens. The penetration of crushed ballast into subgrade soil significantly increases the porosity of subgrade soil, particularly at the interface between ballast and subgrade. This increase in porosity is more pronounced with larger average grain sizes of ballast. At the mesoscopic scale, larger average grain sizes of ballast lead to more localized high contact forces and more significant stress concentrations. The parallel gradation method substantially affects the mechanical properties of ballast penetration into subgrade soil, at both macroscopic and mesoscopic scales. Therefore, a cautious approach is necessary when relying on this method for precise assessments.

The penetration of ballast in ballast track significantly affects subgrade performance. A unit specimen was designed with ballast on top and subgrade soil below. Laboratory dynamic triaxial tests and discrete element method (DEM) simulations were used to study the macroscopic deformation behavior and local deformation characteristics of crushed ballast penetration into soil subgrade under dynamic loads. The results indicate that, under train-induced dynamic loads, the ballast and subgrade soil only transmit stress through a limited number of discrete contacts at the interface. As the dynamic stress amplitude increases, the depth of ballast penetration into subgrade soil also increases, exhibiting an exponential relationship with the dynamic stress. The deformation process of the ballast penetration specimens can be divided into three stages: localized compression, shear band formation, and shear band development. Under train-induced loads, ballast penetration significantly increases the porosity of soil samples near the ballast-subgrade interface, and causes significant lateral deformation at the contact interface. Saturated specimens with higher porosity can experience mud pumping under relatively low dynamic stress. The increase in subgrade surface porosity caused by ballast penetration is a significant factor contributing to mud pumping in existing railways. Prevention of mud pumping should focus on preventing the local increase in subgrade porosity caused by ballast penetration. The findings deepen our understanding of the ballast penetration phenomenon and the resulting deformation behavior of the subgrade surface.