The problem addressed in this thesis is the dynamic behavior of end-frame railway bridgessubjected to high-speed train loading, with a particular focus on soil-structure interaction(SSI). The Swedish Transport Administration (Trafikverket) has high demands for suchbridges due to the dynamic loads imposed by trains traveling at speeds up to 320 km/h.

Addressing this issue involved extensive use of Python to develop a parametric spaceincluding bridge parameters such as span length, beam height, cantilever length, endframeheight, end-frame thickness, bearing pad width, and haunch dimensions. The initialdata set included a number of bridges that were not feasible from a static point of view.In order to filter them out, a 2n-factorial experiment was conducted, resulting in a datasetof different end-frame bridge configurations to evaluate their impact on both static anddynamic responses. Using this dataset, a static analysis was conducted to obtain sectionforces such as bending moments, shear forces, and displacements. The results were thenused to filter out non-buildable bridge designs, validated through Eurocode standards.

The dynamic response of these statically buildable bridges from the dataset wasinvestigated using a High-Speed Load Model (HSLM) module from Tyréns, simulatingthe dynamic effect on the bridge structures. Given the large dataset with discretizedend-frame bridges, significant computational resources were required. Therefore, theHSLM analysis was conducted on a supercomputer at the KTH/Parallel Data Centre(PDC) to obtain efficient computation. This analysis yielded kinematic quantities suchas accelerations, rotations, and displacements, which were checked against Eurocodestandards.

Key factors influencing the structural performance of these bridges were identifiedthrough a correlation study of the bridge parameters against the kinematic responses. Thiscorrelation analysis assumed linear/non-linear relationships between the parameters andthe kinematic variables. Additionally, an analysis of the FRF’s used in the computationswas also conducted to confirm the findings from the correlation study and do understandthe dynamic properties of the model, as well as to validate that the HSLM-analysis wascorrectly implemented in the parametric space.

The study revealed that increasing the end-frame length and thickness significantlyreduces acceleration, with additional benefits observed from incorporating wingwalls.Raising the beam height effectively lowers the dynamic response amplitude, though itmay pose challenges for environmental impact and static stability, when regarding thevertical deflection of the main beam. Increasing the span length reduces accelerationbut increases displacement and rotation, while shortening the console length decreasesthe response amplitude across all components. The influence of haunches on dynamicbehavior remains unclear, requiring further investigation to determine their effectivenesswithin the dynamic parameter space.