Abstract
Majority of the railway bridges built in the UK and around the world before the middle of 20th century are of riveted type made of wrought iron or old mild steel. Many of these bridges are approaching their useful design life. Despite their old age, many are still in operation. In-service fatigue cracking and failure in the riveted double-angle connections in such bridges, especially in the rivets and outstanding leg of the angles, had been reported in the past. These connections are one of the most fatigue-prone details found in riveted railway bridges. Fatigue assessment of riveted details is generally based on a global life assessment, such as the S-N approach. The S-N approach involves the challenge of selecting an appropriate detail classification on an S-N basis as well as defining a nominal stress, free from stress raiser effects. By contrast, local stress methods, such as the Theory of Critical Distances (TCD), which directly considers the effect of stress raisers and loading conditions through finite element analysis, may provide a more favourable option for fatigue evaluation since it avoids the need for S-N classification and nominal stress calculation. In this thesis, the fatigue life prediction capability of the TCD and the S-N methods when applied to riveted bridge details was investigated through finite element analysis of simple and complex riveted details focusing on stringer-to-floor-beam connections. The fatigue life predictions of both the TCD and the traditional S-N methods were compared with the experimental data. Thus, a database of static and fatigue tests available in the literature on structural wrought iron and mild steel riveted details and connections was created. The choice of the suitable modelling techniques for the finite element analysis of the riveted details was verified using a benchmark study on a riveted single lap joint. The critical length vs. number of cycles to failure relationship, L vs. Nf, necessary for fatigue analysis based on the TCD was calibrated for structural wrought-iron material obtained from a real bridge. Fatigue assessment of the investigated riveted details was performed based on both the TCD and the S-N methods to quantify the differences. The results of this study showed that the TCD was successful in predicting fatigue life with the predictions falling within the constant amplitude scatter bands of the experimental data. In the single lap joint, the TCD method accurately estimated the average rivet clamping force values developed in the rivets of the specimens experimentally investigated. In the butt joints, the results of the TCD method was found to conformed well with the experimental research in the literature by predicting higher rivet clamping forces in the specimens with longer grip length. The novel formalisations of the TCD effectively predicted the fatigue life of full-scale riveted built-up girders with the results in the medium- and high-cycle always falling inside the CA scatter bands of the experimental data. The accuracy of the TCD method in estimating the rivet clamping forces present in the rivets of the investigated girders was found to be very high. The TCD was highly accurate when used to estimate the fatigue life of different components of the stringer-to-floor-beam connection subjected to four-point bending with the predictions falling in the scatter of the available experimental data. The hotspot locations identified by the TCD method were consistent with the results of the experimental study. By contrast, the predictions of the S-N method were very nonconservative in the case of the stringer-to-floor-beam connections (up to a factor of 2900). In general, the S-N method predictions were found to be sensitive to the choice of the detail classification. In the case of the stringer-to-floor-beam connection, Modified Class B resulted in the most nonconservative results (up to a factor of 10) when compared with the results of the other S-N classifications. The findings of this thesis may provide the bridge owners and authorities with a safe and effective alternative method in determining the remaining fatigue life of such bridge details.