The ballastless track system (BTS) is widely adopted in the construction of high-speed railways (HSRs), in which the vehicle-induced cyclic stresses are evenly transferred to subgrade at a smaller amplitude compared with the conventional ballast track system. In this study, both static and cyclic triaxial tests were performed on a subgrade fill procured from a HSR construction site in Mainland China. The subgrade fill contained high fine content (d < 0.075 mm). The stress-strain-strength relationship from static tests provided a reference for the cyclic tests. The permanent axial strain and resilient modulus under cyclic HSR loading were studied based on the stress levels in the field. Based on the test data, a new empirical equation is proposed to predict the accumulated permanent axial strain with cycle number and low cyclic deviator stress. A new method for determining all parameters in this equation is proposed and verified by fitting test data. The results show that the predictions from the newly proposed relationship are in good agreement with the experimental results. The resilient behavior of the fill under cyclic loading is also investigated. It is found that under low-stress levels the resilient modulus slightly decreases with the increase of stresses, which is not well described by existing models.

Estimating the permanent strain of granular materials in embankments subjected to cyclic loading is a major challenge for transport engineering projects. In practice, the development of permanent strain can be divided into two periods, postcompaction and secondary cyclic compression. In this study, the existing literature on prediction of permanent strain is reviewed in detail, and the key advantages and limitations of each model are presented and discussed. Based on the gaps identified in the existing literature, a new model for predicting the permanent strain of granular materials under low cyclic loadings is proposed. This model defines two new terms, “representative cycle number” and “reference strain line,” to distinguish the postcompaction and secondary cyclic compression periods accurately. Specifically, the new model correlates the stress states, first with the accumulated strain at the end of the postcompaction period, and then with the strain rate in the secondary cyclic compression period, with good accuracy. This model eliminates the requirement for static compression tests, which are normally needed for the existing models. The new model also avoids the determination of resilient modulus, which is not a competent definitive parameter for evaluating the performance of granular materials. Further, the new model is validated by predicting the permanent strain development of two types of granular materials that are adopted in pavement subgrade and railway subgrade, respectively, in cyclic triaxial tests. The new model is applied in a trial for predicting the long-term settlement of a full-scale physical model of railway embankment under cyclic loading. The results indicate that the new model can effectively capture and accurately predict the permanent strain of testing materials under various stress states and testing conditions. It is also proved that the practical value of the new model is promising. The effects of moisture content, particle size distribution, and compaction degree were not considered, and further studies are recommended to investigate these factors.