Evaluation of fatigue properties using estimation methods for puddled iron (pre-1900) and mild steel (post-1900)

Purpose This research aims to determine and evaluate fatigue properties using estimation methods based on the monotonic tensile strength properties of old metallic bridge materials – puddled iron (pre-1900) and mild steel (post-1900) – and to rank the estimation methods. Design/methodology/approach Fatigue properties can be estimated using monotonic tensile strength data. In this study, 10 estimation methods were employed: Manson's universal slopes (1965), four-point correlation (1965), Mitchell's method (1977, 1979), modified universal slopes (1988), uniform material law (1993), modified four-point correlation (1993), modified Mitchell's method (1996), Meggiolaro's median (2004), modified Park–Song's method (2018) and GP-symbolic regression (2024). This investigation uses the monotonic tensile strength properties of old metallic bridge materials (Eiffel, Luiz I, Fão, Pinhão, Trezói and Várzeas) to evaluate estimation methods. A comparison and discussion are made between the known experimental and estimated strain-life data. Findings The Mitchell's, modified Mitchell's and Manson's universal slopes methods consistently provide accurate and robust estimates of fatigue parameters for puddled iron (pre-1900) and mild steel (post-1900). The modified Park–Song method is effective only for Luiz I and for puddled iron (pre-1900), but it lacks robustness in general. The modified four-point correlation and GP-symbolic regression methods show poor agreement with experimental fatigue data and fatigue curves for the old metallic materials, especially in the high- and low-cycle regimes. The fatigue properties of the puddled iron (pre-1900) and the mild steel (post-1900) were estimated to serve as reference mean values. Originality/value This research contributes to the quality evaluation and ranking of estimation methods used to predict the fatigue properties of old metallic bridge materials (pre-1900 and post-1900) based on their monotonic strength properties. In many cases, only the monotonic strength behaviour of these materials is known. This makes it essential to evaluate and develop formulations that allow the identification of relationships between fatigue resistance and monotonic strength properties.