Railway Ballast: Grain Shape Characterization to Study its Influence on the Mechanical Behaviour

As several studies showed, mechanical performance of ballast depends greatly on its size, mineralogy and shape. Unlike the formers, particle shape remains a poorly specified property. The main objective of the study is then to first characterize the shape of ballast, then to understand its impact on the mechanical performance, by means of numerical simulations using the Discrete Elements Method (DEM), performed on LMGC90 software, with ballast grains represented as irregular polyhedrons. Starting from a description obtained experimentally by 3D digitization of the particle surfaces as point clouds, we will present a method to characterize and generate a set of virtual grains that are morphologically representative of real ballast grains. The model relies on a statistical modelling of the ballast grain morphology based on a dimensionality reduction approach leading to an optimal and nearly exhaustive shape characterization. In this paper, we will focus on the efficiency of both characterizing and generating methods and describe their advantages, as well as an outlook for numerical simulations.