Abstract This work presents methodologies for the examination of uncertainties in plant loading conditions for incorporation in probabilistic creep-fatigue (CF) crack initiation assessments. These approaches are based on the use of plant data for transient (TR) as well as steady-operating (SO) conditions, with both potentially having large contributions towards creep-fatigue damage. Conventionally, the stress-states in a boiler plant component are found using thermal and mechanical (elastic) finite element (FE) models. The main inputs to these models are the boiler steam temperatures, while the outputs are the six stress components (SCs) and the metal temperature (MT) at the assessment location(s) of interest. The proposed methodologies were developed based on experience gained from examining a tube-plate (TP) plant component, for which historical data was available. In a probabilistic lifetime assessment, the aim of these approaches is to replace time intensive FE runs with probabilistic alternatives which incorporate the load variabilities of interest. The value of these approaches lies in the avoidance of running FE models for every probabilistic trial (e.g. a Monte-Carlo simulation may require more then 10 5 trials), which may be computationally prohibitive.

中文翻译：

---

工厂负载不确定性及其在概率蠕变损伤评估中的结合

摘要 这项工作提出了检查设备加载条件不确定性的方法，以纳入概率蠕变疲劳 (CF) 裂纹萌生评估。这些方法基于对瞬态 (TR) 和稳定运行 (SO) 条件的工厂数据的使用，两者都可能对蠕变疲劳损坏有很大贡献。传统上，锅炉设备组件中的应力状态是使用热和机械（弹性）有限元 (FE) 模型找到的。这些模型的主要输入是锅炉蒸汽温度，而输出是感兴趣的评估位置的六个应力分量 (SC) 和金属温度 (MT)。建议的方法是根据从检查管板 (TP) 设备组件中获得的经验开发的，哪些历史数据是可用的。在概率寿命评估中，这些方法的目的是用包含感兴趣的负载可变性的概率替代方案替换时间密集型有限元运行。这些方法的价值在于避免为每个概率试验运行有限元模型（例如，蒙特卡罗模拟可能需要超过 10 5 次试验），这在计算上可能会令人望而却步。