Klara Mattsson presents her master’s thesis

Student: Klara Mattsson
Main Supervisor: Jens Nielsen
Examiner: Anders Ekberg
Opponent: Jaseung Lee

Abstract of thesis

The railway system is reliant on trains running according to the train schedule to avoid delays, which is why there need to be as few interruptions as possible. This thesis focuses on wheel tread damage in the form of wheel flats, which can cause big interruptions in the Swedish railway system since if their length is longer than 60 mm, the wagon immediately needs to be taken out of regular service. The wagons are then put aside on, for example, an extra line on a station, which could have been used for oncoming trains. This thesis aims to understand the wheel-rail impact loads caused by wheel flats, and how varying variables such as the length of the flat, axle load, velocity, time, and unsprung mass influences the load.
The thesis is divided into two main areas: analysis of data from wheel impact load detectors and simulations. The analysis is performed on data collected from wheels that have been replaced due to wheel flats, and by going back in history their reg- istered values in wheel impact load detectors were obtained. The simulations were carried out on wheel flats that had been 3D-scanned and were performed so that different axle loads, velocities, and unsprung masses were taken into consideration.
No data from the analysis based on values from wheel impact load detectors showed a value exceeding 350 kN, which is the value set by Trafikverket when a wagon needs to be taken out of regular service. The analysis showed that the values from detectors are not able to show a clear correlation between the resulting impact load and the variables velocity and flat length. This can however be due to reasons such as what lateral position of the flat hit the detector or the detector’s ability to measure peak loads. The conclusion drawn from the analysis is that the values from the detectors are a good indication of the contact forces, but do include a lot of uncertainties. The simulations were performed on two wheels with wheel flats, with the lengths 75 mm and 120 mm. The wheel flat of 75 mm had a depth of 1.4 mm and a force over 350 kN was obtained for 25 tonnes axle load at 100 km/h. The simulations were performed so that the wheel flat could hit the sleeper bay at different positions, since as expected, the most severe case was on top of a sleeper. The 120 mm long wheel flat had a depth of 1.2 mm and the simulated forces were lower than expected. The highest contact force obtained was 240 kN at 25 tonnes axle load and train speed of 140 km/h.
An analysis of the condition of the wheel impact load detectors used in the analysis was also carried out. The analysis studied the detector’s ability to measure static (mean) loads. This showed that the detector’s condition to estimate forces accu- rately varies. Both the calibration of the detectors and the track stiffness at the detector site is very likely to influence the results. In addition, three wheel flats were followed over time, and their values in the wheel impact load detector showed big variations.
The thesis found, just like previous studied work in the area, that the depth of the flat is what is most important rather than the length. From the analysis and the simulations, the wheel-rail impact loads were found to be lower than expected. From the founding of this report, it seems more appropriate to base a decision about a wagon’s continued operation on which axle load the wagon has in combination with the flat length. The best scenario would be if the decision was based on the depth of the flat, however, measuring the depth is hard to apply in reality.