Julian Stow

Statistical Modelling of Wear and Damage Trajectories of Railway Wheelsets

This paper discusses the use of Linear Mixed Models (LMM) and Generalized Linear Mixed Models (GLMM) to predict the wear and damage trajectories of railway wheelsets for a fleet of modern multiple unit trains. The wear trajectory is described by the evolution of the wheel flange thickness, the flange height and the tread diameter; whereas the damage trajectory is assessed through the probabilities of various types of wheel tread damage such as rolling contact fatigue, wheel flats and cavities occurring. Different model specifications are compared based on an information criterion.

Assessing the efficiency of maintenance operators: a case study of turning railway wheelsets on an under-floor wheel lathe

This article assesses the technical efficiency of different operators turning railway wheelsets on an under-floor wheel lathe. This type of lathe is a computer numerical control machine used to turn wheelsets in situ on the train. As railway wheels are turned, a certain amount of the wheel diameter is lost to restore the tread profile and full flange thickness of the wheel. The technical efficiencies of the different wheel lathe operators are assessed using a stochastic frontier analysis, while controlling for other explaining variables such as the flange thickness and the occurrence of rolling contact fatigue defects, wheel flats and cavities. Different model specifications for the stochastic frontier analysis are compared with linear mixed model specifications, showing that the stochastic frontier analysis model exhibits a better Akaike information criterion.

Improving the understanding of SPAD risks usingred aspect approach data

Abstract This paper describes a novel technique for estimating the frequency with which trains approach signals showing a red aspect. This knowledge is potentially important for understanding the likelihood of a signal being passed at danger (SPAD) at individual signals and also for normalisation of SPAD data, both locally and nationally, for trending and benchmarking. The industry currently uses estimates for the number of red aspect approaches based on driver surveys, which are considered to have significant shortcomings. Data for this analysis is sourced from publicly available live feeds provided by Network Rail which give information on train movements and signal states. The development of the analysis model is described and a case study presented. The case proves that there are large variations in the red approach rates between individual signals. SPAD risk assessment at individual signals may be significantly enhanced by the ability to estimate red approach rates for individual signals using the techniques described.

Assessing the potential cost savings of introducing the maintenance option of ‘Economic Tyre Turning’ in Great Britain railway wheelsets

This paper assesses the potential cost savings of introducing a maintenance option known as ‘Economic Tyre Turning’ (ETT) in railway wheelset maintenance in Great Britain. It first develops a life-cycle cost model and puts forward a Monte Carlo simulation procedure to assess the life-cycle costs of different maintenance strategies, including ETT. This Monte Carlo simulation procedure samples from statistical degradation models that estimate the evolution of wear and damage trajectories of different wheelsets, and the maintenance impact of wheel turning in the loss of diameter in a more realistic manner by controlling random effects related to unit, vehicle and month of measurement. The main findings suggest that ETT may provide potential savings of around 0.8% up to 4.4% when compared to a simple wheelset renewal strategy, and between 2.0% and 4.7% cost savings when ETT is used in association with more complex strategies.

Use of railway wheel wear and damage predictions tools to improve maintenance efficiency through the use of economic tyre turning

This paper investigates the wear rate and pattern for wheels turned with thin flanges using economic tyre turning. Economic tyre turning refers to the process of turning wheels to a profile that has the same tread shape but a thinner flange than the design case profile, allowing less material to be removed from the wheel diameter during re-profiling. Modern wheel lathes are typically capable of turning such profiles but the GB railway group standards do not currently permit their use. The paper demonstrates how the wheel profile damage model (WPDM) can be used, with a good degree of accuracy, to predict both the magnitude of wheel wear and the worn profile shape of the design and economic tyre turning re-profiled wheels for service mileages exceeding 100,000 miles. The WPDM simulations were run for two typical electric multiple units (one suburban and one intercity train fleet) and a two-axle freight wagon. Additionally, it discusses the calibration methodology used to adjust the wear coefficients contained within the Archard wear model to improve the accuracy of the WPDM simulation results for specific routes and vehicle types. Furthermore, this paper presents the findings of a trial of economic tyre turning on a fleet of intercity trains. The analysis is extended to predict the effect of using economic tyre turning on rail rolling contact fatigue for typical routes and operating conditions using a series of vehicle dynamic simulations. The analysis considers new 56E1 and 60E2 rails together with a selection of worn wheel. The research provides valuable evidence to support a future change to the standards which will allow train operators/maintainers to implement economic tyre turning policies.

Modelling and simulation of the train brake system in low adhesion conditions

This paper describes the current version of the Low Adhesion Braking Dynamic Optimisation for Rolling Stock (LABRADOR) simulation tool that can predict the train brake system performance and support decision-making in the design and optimisation of the braking system including wheel slide protection, sanders and the blending and control of friction and dynamic brakes in low adhesion conditions. The model has been developed in MATLAB/Simulink and is intended to mimic the braking performance of both older and newer generations of multiple unit passenger trains. LABRADOR models have been initially validated by comparing simulation results for a single car train (Class 153) and two-car train (Class 158) in dry conditions with experimental tests, for tare and crush laden vehicles. This project is supported by RSSB and a technical steering group composed of railway braking experts, suppliers and train operators and manufacturers.