Evaluation of ballast compactness during the tamping process by using an image-based 3D discrete element method

We investigated the dynamic behavior of a railway ballast layer during tamping operation by 3D Discrete Element simulations. The ballast layer was prepared with ballast grains that had a realistic shape and was measured by a laser scanner and modeled by clumped spheres using the dynamic optimization technique. Three types of sleeper models that were embedded in the ballast layer were prepared, and a series of tamping operations (lifting of the sleeper, insertion of vibrating tines and packing, pulling out of tines, and settling of the sleeper) were simulated in a realistic manner. The compaction underneath the sleeper mainly occurred during the insertion of tines and the packing process, while the tine insertion zone was eventually loosened after the tines were pulled out. The final porosity differs with different sleeper models and packing process periods. On the other hand, the coordination number (the number of inter-granular contacts per grain) drastically decreased during the insertion of the vibrating tines and the packing process; then, it recovered to the original value after the tines were pulled out. This change is important to identify the extent of ballast grain agitation due to the processes. The contact force network is sensitive to the applied loading and the place where the tines are inserted, and the resulting force network localization at the final state can be understood by the Coulomb's earth pressure theory. The degree of heterogeneity may be the key to the long-term stability of the ballast layer under repeated train wheel loading.