Influence of load distribution on trough bridges

Sammanfattning: There are approximately 4000 railway bridges in Sweden and a common construction type is the short span concrete trough bridge. With the current standards the load distribution through ballast is assumed to be uniformly distributed with a distribution slope of 2:1 according to the Swedish Administration of Transport or 4:1 according to Eurocode 1. Previous research shows that there are a lot of factors that affects the load distribution through the ballast and that the distribution rarely is uniform. Different load patterns on bridges can result in different responses in the structure and it is possible that a more optimized evaluation of the loads could reduce the internal stresses in the bridge. There are gaps in the current literature regarding the structural response to different load patterns on reinforced concretetrough bridges and this master thesis aims to further the research in this area. This report will consist of a literature study where load distribution in ballast is researched in order to find what different load distributions are common and how different parameters affects the load distribution through the ballast. Further, a non-linear FE-model of a typical trough bridge in Sweden that was located in Lautajokki will be developed using ATENA Science. The model will be complete with ballast, sleepers and rails and will be calibrated using the results from a previous full-scale test on the Lautajokki bridge. Four more models will be developed without ballast, sleepers and ballast where the load distribution instead is modelled directly on top of the slab of the bridge. These models will be compared to the model with ballast, sleepers and rail (called the Full model) to see what load distribution that is the closest to reality and how the behavior of the bridge changes depending on the assumed load distribution. The parameters that will be tested and compared during this master thesis is the maximum load capacity, the stiffness, the crack patterns, the stresses in the reinforcement, the moments and shear forces. The load distributions that are tested in this thesis is the Swedish standard, TDOK 2013:0267 (Trafikverket, 2019), the European standard Eurocode 1 (CEN-1991, 2003), a load distribution that is theoretical according to research done by Andersson (2020) (called Realistic load case), and one where the load is assumed to be partially uniformly distributed under the rail seats under a sleeper according to AREMA (2010) (called Partially distributed). The results showed that the realistic load case was the one that was the closest to the Full model since it was the closest load distribution to the Full model for the stiffness of the bridge, the maximum load capacity, the max stress in the reinforcement and the average shear force in the bridge. The only parameters where it was not the closest was for the maximum strain in the concrete and for the average moment in the bridge. This load distribution is however not realistic to use for designing bridges since the pressure distribution is so unnecessarily complex. When it comes to the Swedish standard it also followed the behavior of the Full model closely, it had capacities that were generally larger compared to the Full model, the only exception was the max axle load where it had 1.5% lower capacity. The Swedish standard was also the second closest to the Full model in all tested parameters except for the stiffness. Furthest from the Full model was the load distribution after Eurocode 1 which had the furthest values from the Full model in every tested parameter except for the average moment distribution in the bridge. Eurocode 1 also had lower capacities compared to the Full model for every tested parameterwhich means that this model probably underestimates the capacity of the bridge. The stiffness of this model was however one of the closest to the Full model. The Partially distributed load case had higher capacities compared to the Full model in every measurement. It also had a stiffness that was the stiffest for every measuring point compared to any other load case. This model can probably overestimate the capacity of the bridge. Since non-linear analyses takes a long time to perform linear analyses are more often used to design structures. To test how big the differences are between non-linear and linear analyses all load distribution models will also be run with linear elastic materials to compare the two FEM methods. The comparison between the non-linear analysis and the linear analysis showed that the linear elastic analyses give larger extreme values for both the moments and shear forces which is reassuring since this means that these values are on the safe side. The one exception is the transversal moments for the slab were the moments at the connection to the beam was greater for the non-linear analyses compared to the linear one