Sand particle entrainment and its effects on the wheel/rail interface

Low adhesion in the wheel/rail contact has long been a problem for the rail industry in Great Britain as it causes significant scheduling and safety issues; this is especially true in Autumn. Applying sand to the wheel/rail contact has long been used to mitigate against low adhesion, however there is not a common consensus on what makes a “good” particle for restoring adhesion.
The aim of this project was to optimise the sanding process by investigating what constitutes an ideal adhesion restoring particle. The structure of this investigation included characterising a range of particles and assessing their effect in a small scale tribological test, namely the high pressure torsion rig. Additionally, work was undertaken to study the entrainment of particles into the wheel/rail contact; this was achieved through particle characterisation and the use of a scaled wheel test set-up. Finally, the extended creep force model was parameterised using data accrued from high pressure torsion testing to predict full scale wheel/rail behaviour when sand is present; these predictions were validated using data from a full scale wheel/rail rig.
A literature review was conducted to assess previous sanding research, this resulted in the identification of knowledge gaps in the understanding of the sanding process. The identified knowledge gaps included: the effect different particle characteristics have on wheel/rail adhesion; the mechanisms surrounding the restoration of adhesion by particles; and modelling the effect of particles with regards to their full scale behaviour.
A particle characterisation framework was formulated based off key particle properties for sanding that were identified during the literature review. A range of particles from the international rail industry and other marketed abrasive products were assessed using the particle characterisation framework and the measured data was used in later chapters to compare the importance of different characteristics on adhesion and damage during high pressure torsion testing.
High pressure torsion (HPT) testing has previously been used to simulate the wheel/rail contact by using a rail and wheel specimen to create a flat, annular contact under high normal pressures with the two specimens then rotated over one another and the torque required to do this measured. As part of this project, the test methodology was adapted for testing with granular material and with low adhesion conditions present. An ordinary least squares model was used to assess which particle characteristics had a statistically significant effect on traction and damage. Through this, harder and more circular particles were found to increase traction in dry and wet conditions, whilst in leaf contaminated conditions harder, and less circular particles of an optimum size increased traction. With regards to damage, less circular particles produced rougher surfaces, with larger particles creating more roughness on the surface of the wheel specimen surface.
Further HPT tests were conducted to assess the effects of multiple wheel passes and varying normal pressure on traction using one type of rail sand. It was observed that in dry and wet conditions multiple passes had little effect on traction, whereas in leaf contaminated conditions traction decreased. In dry and wet conditions, increasing normal pressure saw an increase in traction, in contrast with leaf contaminated conditions where an increase in normal pressure resulted in a decrease in traction.
Scaled wheel tests were performed to corroborate findings concerning particle entrainment observed during particle characterisation and HPT testing. In dry conditions, much of the sand particle was expelled out of the contact upon being crushed. In comparison, wet conditions saw a greater amount of the particle being retained in the contact.
Finally, HPT data was used to parameterise the extended creep force (ECF) model. Good agreement was seen between the HPT data and ECF model, indicating the parameterisation process was successful. The parameterised ECF model was then used to generate predicted creep curves for a typical real world wheel/rail contact; in dry, sanded conditions, adhesion was estimated to be between 0.4-0.5; in wet, sanded conditions, between 0.3-0.4; and in sanded, leaf contaminated conditions, between 0.1-0.2.
Full-scale validation of the ECF predictions was hampered by unusually low traction measurements on the FSR, however the trends between contact conditions and normal loads were similar to the ECF predictions. In addition, field data from unsanded and sanded, leaf contaminated conditions matched up very well with ECF predictions.