Modelling of flow-induced loading and fatigue crack trajectory in pipelines

流动引起的结构效应，如应力、振动、疲劳和裂纹扩展，对加拿大能源工业核心管道的运营和维护构成重大风险。在地下埋管存在弯曲的情况下，预测这些特征的结构影响变得更具挑战性。拟议的研发工作旨在开发一种基于机器学习的模型，该模型可识别各种操作条件下管道中的流动诱导结构特征。建立了一个跨学科和多样化的研究小组，以研究这一现象的各个方面。最终目标是开发一个综合预测模型，以表征和预测管道中流动诱导结构伪影的形成和轨迹，流速、流体粘度、管道内径、弯曲角度和表面材料。该研发项目由三个综合主题组成，这些主题将联合收割机的实验、计算和理论方法与各自的活动和时间表相结合。 在主题I中，流体力学专家评估管道中流动引起的负载的物理特性，并在一系列流动条件和流体特性中开发振动，应力和疲劳的半经验模型。同时，专题II的结构专家研究管道中结构裂缝、疲劳和应力的形成、传播和轨迹。最后，在主题III中，将结果集成到一个新的基于机器学习的模型中，以使用从常规管道监测过程中获得的流量数据来预测管道中结构特征的形成和传播。该项目有助于推进对管道流动动力学，流动诱导载荷，疲劳裂纹扩展和基于机器学习的物理模型的基本理解，同时还开发了一种改进管道运营和维护的新工具。该项目通过提高管道的安全性和效率并减少其环境足迹，对加拿大能源行业产生了巨大的影响。此外，该项目使HQP培训成为加拿大未来的领导者。

Flow-induced structural effects, such as stress, vibration, fatigue, and crack-propagation, pose a major risk in operations and maintenance of pipelines at the heart of Canada's energy industry. Predicting structural impact of such features become more challenging in the presence of bends for pipes buried underground. The proposed R&D efforts aim at developing a machine-learning-based model that identifies flow-induced structural features in pipes at various operating conditions. An inter-disciplinary and diverse team of researchers are established to look at various aspects of this phenomenon. The final objective is to develop a comprehensive predictive model to characterize and forecast the formation and trajectory of flow-induced structural artifacts in pipes across a broad parameter space, e.g., flow speed, fluid viscosity, pipe inner diameter, bend angles, and surface material. This R&D project consists of three integrated themes that combine experimental, computational, and theoretical methodologies with their individual activities and timelines. In Theme I, fluid mechanics experts evaluate the physics of flow-induced loading in pipes, and develop semi-empirical models for vibration, stress, and fatigue, across a range of flow conditions and fluid properties. In parallel, structural experts in Theme II study the formation, propagation and trajectory of structural cracks, fatigue, and stresses in pipes. Finally in Theme III, the results are integrated within a novel machine-learning-based model to use flow data obtained from routine pipe monitoring processes to predict the formation and propagation of structural features in pipes. This project contributes to advancing knowledge in fundamental understanding of pipe flow dynamics, flow-induced loading, fatigue crack-propagation, and machine-learning-based physical models, while also developing a novel tool for bettering the operations and maintenance of pipelines. The impact of this project on Canada's energy industry is enormous by enhancing the safety and efficiency of pipelines and reducing their environmental footprint. Moreover, this project enables training of HQP to become future leaders in Canada.