Towards improved regulations and procedures for damaged wheels - Assessment of wheel impact load detectors

Report, 2024

To increase passenger and freight traffic, railway transportation needs to be safe, reliable, environmentally friendly, cost-efficient, and on time. To this end, railway wheels that are not perfectly round because of some type of mechanical damage on part of the running surface (wheel tread) should be avoided as they may lead to severe loading and damage of the track. Resulting infrastructure failures, such as rail breaks, could cause severe traffic delays, derailment, and in the worst case, fatalities.

This is the reason why Trafikverket uses wayside wheel impact load detectors (WILD) in their network to measure the dynamic load generated by each passing wheel. Besides a limit on maximum allowed peak load, Trafikverket imposes regulations on the maximum allowed circumferential length of a wheel tread damage. The train must be stopped for visual inspection if the measured peak load exceeds the alarm limit 350 kN. If the wheel tread damage exceeds 60 mm in length, the vehicle with the damaged wheel must be taken out from the train independent of train speed and axle load. The stopped train (or wagon) might then block one track that should have been used for passing traffic. The resulting delay and required repair of this train leads to primary costs for the infrastructure owner (Trafikverket) and rolling stock operators (such as Green Cargo and SJ). Secondary costs in these events are caused by subsequent trains being delayed, inducing costs from reimbursing affected passengers/freight customers etc. This is a considerable issue in the Swedish railways, where a substantial part of the network is only single track.

Thus, regulations for removal of out-of-round wheels have a substantial influence on punctuality and costs. The implementation of WILDs to measure dynamic loads offers the opportunity to define a criterion for removal of wheels based solely on the measured load, thereby also removing the need for manual inspection and increasing employee safety. Overall, it is vital that the detectors have a robust and transparent calibration procedure to ensure accurate measurements of dynamic wheel loads, and that they are subjected to regular maintenance and monitoring of track geometry in the detector area.

The present study was carried out in the project Improved regulations and procedures for damaged wheels, which was funded by the Swedish innovation agency VINNOVA and performed 2023-07-01 – 2026-06-30 by representatives from Trafikverket, Chalmers University, Green Cargo and SJ. Parts of the study have been funded within the Horizon-ER-JU-2022-FA5-01 project TRANS4M-R under grant agreement no 101102009. The current report is the final deliverable from work package 1 in the VINNOVA project, and it has the following contents:

- The background to the VINNOVA project is presented in Section 1 of this report.

- In Section 2, the alarm limit and current regulations are summarised.

- The two types of WILD, Schenck and voestalpine zentrac, currently used in the Swedish network are described in Section 3. Trafikverket’s regulations on their placement and calibration are summarised.

- A case study investigating measured data from six Schenck detectors along the route of Stålpendeln is presented in Section 4. In total, 823 detector passings by 149 different wheelsets (with at least one of the wheels on the axle being defective) were included in the analysis. It is shown that no evident correlation between wheel flat length, train speed and peak load could be found. In parallel, for one investigated 75 mm long wheel flat, all measured peak loads and mean loads for two journeys in loaded conditions and three journeys in tare conditions have been studied. A large variation in measured loads between the different detectors is observed, indicating that besides the variation in speed and other influencing variables, the condition of the detectors might have affected the measured loads. The accuracy of the measured dynamic loads is unknown. Field measurements indicate that irregularities in track geometry and track stiffness might have contributed to the scatter in measured peak loads demonstrating the importance of regular monitoring of the conditions at the detectors.

- The consequences of out-of-round wheels in terms of rail damage and costs for Trafikverket are discussed in Section 5. Trafikverket stores information on the causes and consequences of each reported rail damage. In 2023, no rail breaks due to wheel flats or damaged wheels were reported. Inspections of vehicles or tracks (such as those occurring after alarm/warning levels) caused 1000 hours of delay in 2023, while they caused around 300 hours of delay in 2013.

- Mean loads and dynamic loads measured for two different bogie types used by Green Cargo are compared in Section 6. It is concluded that all studied detectors seem to measure mean loads accurately. However, there seems to be a seasonal variation in dynamic loads measured by the zentrac detectors, particularly for the unloaded vehicles and independent of braking system and bogie type. The reason for this is unknown and needs to be solved to minimise the risk of false alarms and unnecessary costs.

- The action plan used by SJ in case of a detector alarm is described in Section 7. Generally, very few SJ trains are stopped due to their active condition-based maintenance strategy. To specifically monitor wheel degradation and trends, Power BI-reports are generated. Quarterly, Trafikverket and the Railway Undertakers meet in a forum named ‘Industry common management detectors’ (Branschgemensam förvaltning Detektorer). To obtain a common understanding of the situation and aiming to focus on the incidences that lead to the highest negative impact in terms of delay minutes, a Power BI tool named ‘Common situation picture’ (Gemensam lägesbild) has been developed.