Monitoring railway traffic loads using bridge weight-in-motion

While there has been much research into the field of Bridge Weigh-in-Motion (B-WIM), this research has focused on road traffic. Because of the differences in traffic properties, it is not applicable to B-WIM for railway traffic without some modifications. The research contributions accounted for in this licentiate thesis is a set of algorithms for performing B-WIM calculations on railway traffic. It has been implemented in the MATLAB language as a toolbox of functions, tailored to easily performing B-WIM calculations. This toolbox, called Twim, is described in detail in the appendices of this thesis. This thesis describes the algorithms of the B-WIM system. That includes algorithms for determining the speed of a train, an algorithm for determining the influence line of a bridge at a sensor position from the signal of a passing train, and a simple algorithm for locomotive identification. Lastly, this thesis outlines some of the results obtained from measurements using the system, and investigates dynamic effects as well as changes in bridge properties over time. As a result of the research conducted in this thesis, the following conclusions can be drawn: The algorithms described in this thesis are suitable for B-WIM on railway traffic; Different types of bridges can be instrumented for B-WIM by measuring transverse strain instead of longitudinal strain, since the former gives a more local traffic effect. Longer bridge span and lower damping ratio do however significantly reduce the accuracy of the system; Long term effects such as changes in bridge properties due to temperature changes could not be detected during the seven month measurement period; High vertical bridge deck acceleration levels are often detected, but they are found to be the result of wheel defects rather than resonance phenomena; Wheel defects significantly reduce the accuracy of the B-WIM system. (A). http://www.byv.kth.se/publikationer/pdf/raid/LicAvhandling-Axel.pdf