A DEM-FEM Coupling Framework Applied to Railroad Infrastructure Simulations

  • Ullrich, Anita (Fraunhofer-Chalmers Centre)
  • Quist, Johannes (Fraunhofer-Chalmers Centre)
  • Cromvik, Christoffer (Fraunhofer-Chalmers Centre)
  • Jareteg, Klas (IPS Particle Technology AB)
  • Bilock, Adam (IPS Particle Technology AB)
  • Edelvik, Fredrik (Fraunhofer-Chalmers Centre)

Please login to view abstract download link

Swedish and other European governments invest significant resources in railroad infrastructure, including maintenance and construction. The degradation of track ballast layers is one of the most critical maintenance issues. Hence, it is of significant interest for infrastructure owners to find novel solutions to mitigate the problem by improving design and maintenance operations. However, established tools for the simulation of railroad systems typically consider the ballast as a solid continuum structure, while in practice, the discrete nature of the particle assembly has to be accurately represented in the model. The sleepers and rails must be modelled as solid structures, which results in the complex coupled problem of combining particulate and structural analysis models. In this paper, the simulation of railroad infrastructure is performed with an explicit surface coupling algorithm of the Discrete Element Method (DEM) and the Finite Element Method (FEM). The ballast layer is represented by individual convex dilated polyhedral particles in DEM, where the computations are performed on the GPU. The rail system with sleepers and potential sublayers is modelled with solid structures in FEM. Properties of the ballast bed, such as particle shape, size distribution and packing density, are found to have a significant impact on the pressure distribution within the bed and the attenuation of vibrations. Thus, the variance between different ballast bed configurations and load patterns needs to be analysed. The resulting pressure distributions show higher variance over time and less symmetry than results from pure FEM approaches. The results also show that accurate particle shape representation, as well as high computational performance, are critical aspects to achieving predictions on a relevant scale. The ability to study the ballast layer as a particulate system provides a new perspective on dynamics in tracked ballast structures.