Hakimi, Aziz Urahman; (2025) An experimental investigation of the contact behaviour of railway ballast. Doctoral thesis (Ph.D), UCL (University College London).

Preview

Text Hakimi_10211106_thesis_sig-removed.pdf Download (23MB) | Preview

Abstract

Accurate simulation of railway ballast behaviour in discrete element models (DEM) relies on precise contact-scale parameters, such as stiffness, friction, and damping. However, these micromechanical mechanisms are difficult to investigate experimentally and therefore remain poorly understood, especially under varying environmental and loading conditions. In most countries, railway ballast seems to be sourced from nearby locations but comparisons between properties of different ballasts and their performance are rare. This research addresses these gaps by investigating the contact-scale behaviour of ballast particles from multiple geological sources using unique experimental facilities and advanced experimental methods to enable more accurate characterisation of their mechanical and surface properties, allowing an informed comparison between the different types of ballast. Railway ballasts of various origins and mineralogy (andesite from China, basalt from Australia, granites from the UK and China) were studied under varied loading conditions -- monotonic normal and tangential loading, normal cyclic loading and cyclic tangential loading, dry and wet contacts under different rates (slow and fast). The tests were carried out in a custom-built inter-particle apparatus—originally developed by Coop and Senetakis (2014) and later refined by Nardelli et al. (2016) and Wong et al. (2019), which was further modified by the author for the purpose of this study with help from Dr. Saurabh Singh. These modifications included increasing the size and number of vertical bearings and washers to enhance stiffness and control. Comprehensive calibration of load cells, displacement transducers, and the sliding mechanism ensured accurate force and displacement measurements. Particular attention was given to the methodology of carrying out the tests for more accurate results, and new procedures were developed. Among the outcomes are the new ways of applying load, with normal loading applied at a standardised rate of 150 mm/hr to avoid creep and asperity damage. Tangential shearing was also optimised and conducted at 0.1–0.5 mm/hr. High-frequency data from cyclic tests (recorded every 3–7 seconds) was processed using custom MATLAB scripts developed specifically for this research for cleaning, slope correction, and precise determination of tangential stiffness and friction. This new methodology provides a robust base for carrying out particle-to-particle tests and obtaining more reliable results. A new procedure was also developed to quantify the roughness of the ballast surface at the contact and its evolution during loading. A systematic method using the ConfoMap software was developed, in the absence of industry standards. The method includes optimised Z-stack imaging (μm); a cut-off value selected through Power Spectral Density analysis in MATLAB, leading to a standardised 16 μm cut-off; form removal using a 9th-degree polynomial to exclude form from surface height data; optimal roughness extraction method, selecting a 0.2 × 0.2 mm area after form removal for filtering, ensuring consistent and representative measurements. Mechanical testing involved monotonic and cyclic loading of ballast types including andesite (China), basalt (Australia), and granites (UK and China), under dry and wet conditions. Test results show that in monotonic normal loading, the load-displacement response of the contact is initially soft (0.01-0.5 N/μm) and with an increase in the load, the response gets stiffer (5-10 N/μm). The experimental response does not follow predicted responses from existing contact models such as Hertz (1882), and Modified Hertz (Yimsri and Soga, 2001). The monotonic and cyclic shearing behaviour of ballast materials revealed distinct friction and stiffness characteristics. In monotonic tangential loading, the stiffness was found to be the highest at the start (1-3 N/μm, static friction) decreasing gradually to zero (kinetic friction, onset of sliding) at about a tangential displacement of 100-200 μm. The monotonic tangential stiffness increases with an increase in normal load. Post-sliding, the tangential force becomes constant with the coefficient of friction from 0.4-0.9. Andesite exhibited the highest friction (0.65-0.9) due to its rough texture, while Chinese granite showed the lowest (0.15-0.5) owing to its smooth surface. Basalt and fresh granite displayed moderate friction (0.4-0.7), with used granite performing similarly but more consistently. Washed-used granite achieved higher friction (0.7-0.8) and maintained the highest tangential stiffness across all conditions, reflecting improved stability after cleaning. In the cyclic tangential loading, stiffness was found to increase with the number of cycles from (1 to 3 N/μm) as well as the coefficient of friction (0.6-0.8), because of changes in the surface roughness, however, the stiffness decreased as the tangential stiffness increased. Cyclic tests showed that basalt, Chinese granite and washed-used granite had the highest friction coefficients (0.65-0.8). Andesite and used granite exhibited moderate coefficients (0.6), reflecting balanced performance under shearing. Fresh granite, however, had the lowest coefficient (0.5-0.7), indicating reduced frictional properties. Interestingly, despite prior usage, used granite maintained a coefficient comparable to andesite, showcasing its durability and potential for reuse. Chinese granite exhibited the least vertical displacement during cyclic shearing, while fresh granite showed the most, attributed to its angularity. Basalt and used granite resisted deformation well, while washed-used granite showed slightly higher displacement, potentially from washing-induced degradation. The rate of abrasion during cyclic shearing identified Chinese granite and used granite as the most durable, with minimal wear under high loads, while fresh granite and andesite showed significant wear, raising concerns about their longevity. Basalt and washed-used granite consistently exhibited high tangential stiffness, though flooding reduced their performance. Further, the cyclic stiffness and coefficient of friction decrease in the presence of water at contact, however, it was found that normal load level and rate of the cycles do not affect the coefficient of friction. Roughness was measured before and after the test for all types of ballasts, and it was found that the asperities deformed through plastic deformation. This research identifies Chinese granite, basalt, and washed-used granite as the most reliable materials for demanding railway ballast applications due to their superior durability, stability, and frictional performance. Recommendations include prioritising these materials for high-performance applications, optimising washing treatments to enhance material reuse, and advancing modelling frameworks to simulate real-world conditions more accurately. Future work could focus on long-term performance testing, dynamic loading simulations, microstructural analyses, and the development of hybrid ballast materials to improve the sustainability and efficiency of railway infrastructure.

Download activity - last month

Export as CSVXMLJSON

Download activity - last 12 months

Export as XMLCSVJSON

Downloads by country - last 12 months

Export as XMLCSVJSON

Archive Staff Only

View Item