Statistical evaluation of fatigue strength of double shear riveted connections and crack growth rates of materials from old bridges

•Fatigue crack growth (FCG) data from puddled iron specimens was analysed.•A statistical analysis based on linear regression model for the experimental FCG data was made.•Fatigue tests on double shear constructional detail specimens in puddled iron were carried out.•A statistical evaluation of fatigue tests based on the Castillo & Fernández-Canteli model was carried out.•A comparison between experimental fatigue data, statistical models and construction norms is proposed. The maintenance of old riveted bridges has become a major concern for railroad managers across Europe. These metallic structures, built on their majority on the second half of the 19th century, are indeed submitted to ever increasing traffic loads (augmentation in weight and speed of rolling stock). To assess the remaining life of these metallic bridges, some critical structural details have been identified and associated to S-N curves to be used in damage estimation (using Palmgren-Miner's rule for cumulative damage, for example). These constructional details are often described in construction standards, such as the EN 1993-1-9, the BS 5400 and the AASHTO standards. The particularity of older hot-riveted bridges is that their numerous geometric configurations are often under-represented, thus, limiting the accuracy of damage estimation. To compensate the absence of hot riveted assemblies in the standards, numerous research groups have performed extensive experimental campaigns. However, the heterogeneity and rarity of the material (no longer produced nowadays), as well as the very diverse geometrical configurations used at the time of the construction of the bridges (given the novelty proposed by the introduction of metal as a new material in monumental construction), makes difficult the comparison of experimental data. In order to facilitate interpretation and differentiation of fatigue data of a particular type of constructional detail, experimental data from double shear assemblies manufactured from three different metallic ancient bridges is considered (French and Portuguese puddled iron bridges). Then, through statistical analysis (linearized boundaries and the Fernández-Canteli model), the S-N curves corresponding to their structural configuration are identified. Comparisons to the EN 1993-1-9, the BS 5400 and the AASHTO standards are also proposed. The combination of the S-N curves and Fracture Mechanics approaches can be used to evaluate the residual life of the old riveted meta