Abstract This study addresses the fatigue crack growth behaviour of a material from an old riveted steel bridge, the Portuguese Fao Bridge, based on an experimental program of constant and variable amplitude loading tests. The material under consideration is a centenary puddle iron, for which information on fatigue crack growth under variable amplitude loads is not available in the literature. Additionally, the experimental program aimed at analysing the fracture surfaces of the specimens through observations under Scanning Electron Microscopy (SEM), and correlate their characteristics with macroscopic fatigue crack growth behaviour. Furthermore, a crack closure analysis was performed using electrical strain gauges applied on the back face of the CT specimens. The fatigue crack propagation along the parallel to the rolling direction (L) revealed to be higher when compared with the transverse direction (T). The experimental data obtained in this study were compared with existing fatigue crack propagation models, such as the Paris model with linear damage accumulation, according to Miner’s rule, as well as with more complex crack propagation models, with capacity for modelling retardation effects of single overloads. The Paris model with linear damage accumulation, despite its simplicity, led to the best predictions for the generality of variable amplitude random block loading. The original Wheeler model is enough to correlate the crack propagation data under overloads, since the application of more sophisticated models would not be justified by the extremely high scatter observed in these old materials.

中文翻译：

---

变幅加载下fão桥水坑铁疲劳裂纹扩展模型

摘要 本研究基于恒幅和变幅加载试验的实验程序，研究了旧铆接钢桥（葡萄牙 Fao 桥）材料的疲劳裂纹扩展行为。所考虑的材料是百年水坑铁，文献中没有关于可变振幅载荷下疲劳裂纹扩展的信息。此外，该实验计划旨在通过扫描电子显微镜 (SEM) 下的观察来分析试样的断裂表面，并将其特征与宏观疲劳裂纹扩展行为相关联。此外，使用应用于 CT 样品背面的电应变仪进行裂纹闭合分析。与横向 (T) 相比，沿平行于轧制方向 (L) 的疲劳裂纹扩展显示更高。本研究中获得的实验数据与现有的疲劳裂纹扩展模型进行了比较，例如根据 Miner 规则具有线性损伤累积的巴黎模型，以及更复杂的裂纹扩展模型，能够模拟单个过载的延迟效应. 具有线性损伤累积的巴黎模型尽管很简单，但对可变振幅随机块加载的普遍性做出了最好的预测。原始惠勒模型足以关联过载下的裂纹扩展数据，因为在这些旧材料中观察到的极高散射无法证明更复杂模型的应用是合理的。