The development of Discrete Element Model (DEM) of railway ballast for the purpose of studying the behavior of ballast particles during tamping is addressed in a simulation study, with the goal of optimizing the railroad tamping operation. A comprehensive literature review of applicability of DEM techniques in modeling the behavior of railway ballast is presented and its feasibility in studying the fundamental mechanisms that influence the outcome of railroad tamping process is analyzed. A Discrete Element Model of railway ballast is also developed and implemented using a commercially available DEM package: PFC3D. Selection and calibration of ballast parameters, such as inter-particle contact force laws, ballast material properties, and selection of particle shape are represented in detail in the model. Finally, a complete tamping simulation model is constructed with high degree of adjustability to allow control of all process parameters for achieving realistic output. The analysis shows that DEM is a highly valuable tool for studying railroad tamping operation. It has the capability to provide crucial and unprecedented insights into the process, facilitating not only the optimization of current tamping practices, but also the development of novel methods for achieving sustainable improvements in track stability after tamping in the future. Different ways of modeling particle shapes have been evaluated and it has been shown that while using spheres to represent irregular ballast particles in DEM provides immense gains in computational efficiency, spheres cannot intently capture all properties of irregularly shaped particles, and therefore should not be used to model railway ballast particles. Inter-particle and wall-particle contact forces are calculated using Hertzian contact mechanics for determining ballast dynamics during tamping. The results indicate that the model is able to accurately predict properties of granular assemblies of the railway ballast in different test cases. The developed model for simulating tamping operation on a half-track layout is expected to be extended in future studies for evaluating rail track settlement and stability, optimization of tamping process, and performance of different ballast gradations. / MS / Development of a virtual simulation model for the stone bed which forms the foundation of traditional rail track structures is discussed in this study for the purpose of improving a conventional railway maintenance practice called tamping. The stone bed, called ballast, is flexible and is susceptible to undesirable deformation due to the forces from train traffic on the rail tracks over their service time. Therefore, periodic restoration of track structure is performed by tamping to maintain the operational quality of the rail tracks and reduce the risk of train accidents. This simulation model is intended to accelerate the scientific development of the current tamping practices by providing unprecedented insight into the behavior of small stones which form the bulk of the ballast and obviating the requirement for costly physical experimentation. The nuances of the mechanical behavior of ballast have been examined by a comprehensive literature review and the selection of a modeling technique called Discrete Element Modeling (DEM) has been justified for modeling of ballast owing to its suitability in capturing intricate dynamics of ballast stones. The virtual simulation model which is developed as results of this work has been found to be extremely efficient in realistically predicting the outcome of tamping process for any set of conditions of interest. This implies that quality of the rail tracks after tamping can be studied for a variety of different test cases and most optimized set of tamping parameters which results in maximum track quality can be analyzed. However, it was observed that the accuracy of the results obtained from the simulation model is dependent on the level of detail which is used to input properties of the ballast into the model. Low level of detail results in less accurate results whereas a high level of detail takes an unreasonably long time to solve. Therefore, a compromise has to be made between accuracy and solution time while programming the simulation model, and additional work is required in the future to improve the solution speed of the model.