Recommendations for revision of maintenance procedures and regulations for removal of damaged wheels

Report, 2026

The present study was carried out within the project Improved Regulations and Procedures for Damaged Wheels, funded by the Swedish innovation agency VINNOVA and conducted between 1 July 2023 and 30 June 2026 by representatives from Trafikverket, Chalmers University of Technology, Green Cargo, and SJ. Parts of the research were also supported through the European project TRANS4M-R (Horizon Europe Rail Joint Undertaking, Grant Agreement No. 101102009). The overall objective was to improve regulations, maintenance procedures, and operational decision-making related to damaged wheels in order to reduce infrastructure damage, avoid unnecessary train stops, and strengthen collaboration between infrastructure managers and railway undertakings.
The VINNOVA project was divided into four work packages (WPs). This report constitutes the final deliverable from WP4, while also summarising the main findings and implementations from WP1 – Detectors, WP2 – Simulation Models, and WP3 – Track Damage Assessment.
A central finding throughout the project is that the current practice of assessing wheel damage primarily through geometric criteria, particularly wheel flat length, does not adequately reflect the actual risk posed to railway infrastructure or operations. Extensive analyses of Wheel Impact Load Detector (WILD) data, combined with detailed simulations of vehicle–track interaction, demonstrated that wheel–rail impact loads depend on several interacting factors, including wheel-flat depth and shape, axle load, train speed, wheel–rail contact conditions, and local track characteristics. Consequently, wheel defects of similar size can produce substantially different impact loads and resulting effects on infrastructure and vehicles. In several cases, wheel flats exceeding the current 60 mm intervention threshold generated loads well below the alarm level of 350 kN, while shorter but deeper defects produced considerably higher dynamic forces.
Within WP1, existing wheel impact load detectors were critically assessed and detector data from Swedish railway operations were analysed. The work resulted in improved understanding of detector performance and variability, as well as new guidelines for detector placement, maintenance, and calibration that have been adopted by Trafikverket. The studies highlighted the importance of controlling track geometry and track stiffness to obtain a high measurement quality. It also identified the absence of a robust dynamic calibration procedure as an important area for future development.
In WP2, advanced simulation models were used to investigate wheel–rail impact loading and the resulting infrastructure consequences. The simulations demonstrated that the risk of rail break is governed by a combination of dynamic wheel loads, crack size, rail material properties, temperature conditions, and track support stiffness. The analyses further showed that wheel condition and resulting impact-load distributions can be more important than axle load and train speed in determining infrastructure risk. These findings support a transition from purely geometry-based regulations towards risk-based approaches based on measured loads and quantified infrastructure consequences.
Within WP3, methods were developed for using detector data to predict resulting track damage and rail crack growth. The studies demonstrated that detector systems have significant untapped potential for train fleet monitoring and providing support for predictive maintenance planning. Recommendations were developed for future enhancements, including collection of full load histories, implementation of additional detectors in curves, and integration with wheel-profile monitoring systems. Such developments would improve the detection, classification, and management of wheel tread damage and support more reliable condition-based maintenance.
The results from the first three work packages formed the basis for WP4, which focused on implementation and regulatory development. Revised regulations for more reliable track structures were developed and evaluated by Trafikverket. Based on detector data, simulations, operational experience, and international benchmarking, recommendations were made for a more flexible and risk-informed framework for handling wheel-damage alarms. A notable outcome was the implementation of revised speed-reduction rules for vehicles triggering high-level detector alarms, reducing unnecessary operational disruptions while maintaining safety.
The practical benefits of the project are demonstrated through examples from Green Cargo, SJ, and Trafikverket. The studies show that major wheel damage is a relatively rare but high-impact phenomenon. A single detector alarm can result in substantial vehicle downtime, recovery costs, reduced capacity, and extensive traffic delays. For passenger traffic, even short inspection stops can generate thousands of accumulated delay minutes across the network. For freight operators, incidents may result in costs ranging from tens of thousands to hundreds of thousands of Swedish kronor due to disrupted operations, vehicle transport, and asset unavailability. These examples illustrate the value of early detection, reliable monitoring, and risk-based maintenance planning. The removal of the requirement for on-foot track inspections following a high-level alarm is by itself estimated to generate annual socio-economic benefits of more than 50 MSEK through reduced inspection costs, fewer traffic-disruptive interventions, and improved operational efficiency.
Overall, the project demonstrates that a transition from geometry-based assessment of wheel damage towards load-based and risk-based management is both technically feasible and operationally beneficial. The developed methods, updated detector routines, revised regulations, and improved collaboration between stakeholders provide a strong foundation for more effective infrastructure protection, reduced traffic disturbances, lower lifecycle costs, and improved reliability of the Swedish railway system. The project outcomes also provide valuable guidance for future development of intelligent condition-monitoring systems and risk-based railway maintenance practices both in Sweden and internationally.