Numerical investigation of slug flow by smoothed particle method

气液两相管流是石油、化工、核工业等领域的重要流场之一。 这种流动的一个基本困难是预测流型的发生和特性，这与石油管道的安全设计密切相关。由于缺乏可靠的闭合模型，目前的一维计算方法只能取得非常有限的成功，并经常给出错误的预测。本项目的长期目标是为弹状流的一维瞬态模型建立可靠的闭合关系。在为期3年的第一个资助期内，我们采用光滑粒子流体动力学（SPH）方法对弹状流进行了三维大涡模拟（LES）的数值研究，该方法在处理复杂的相间相互作用方面具有很大的优势。本文将在前人工作的基础上，进一步发展适用于气液两相湍流管流模拟的SPH方法。对于典型的构型，我们将研究流体动力学不稳定性的增长、段塞的起始以及起始段塞的进一步发展。我们希望重现弹状流，这是在协议的定性观察和定量测量从实验和研究的启动机制和启动后的动态，导致最终的可持续段塞。在第二个资助期内，我们将专注于通过使用系统化的粗粒度方法来推导一维瞬态模型的闭包。此外，如果时间允许，我们将通过对实验设置和实际管道的广泛模拟来优化导出的闭包。

The gas-liquid pipe flow is of great interest in the petroleum, chemical and nuclear industries due to its frequent occurrence in process equipment and piping systems. A fundamental difficulty for such flow is on predicting the occurrence and characteristics of the flow regimes, which are highly relevant to the safety design of the oil-pipe lines. Due to lack of the reliable closure models, the current one-dimensional computational methods only achieve very limited successes and very often give erroneous predictions. The long-term objective of the current project is to develop reliable closure relations for one-dimensional transient models of slug flow. During the first funding period of 3 years, we carry out numerical investigation of slug flow by three-dimensional large eddy simulation (LES) with the smoothed particle hydrodynamics (SPH) method, which has great advantages on handling complex inter-phase interactions. Based on our previous work, we will develop new SPH methods suitable for simulating gas-liquid turbulent pipe flow. For typical configurations, we will investigate the growth of hydrodynamic instability, the initiation of slugs and the further development of the initiated slugs. We expect to reproduce the slug flow which is in agreement with both qualitative observations and quantitative measurements from experiments and study the mechanism of the initiation and the after-initiation dynamics which leads to the final sustainable slugs. In the second funding period we will focus on deriving the closures for one-dimensional transient models by using a systematic coarse-graining approach. Furthermore, if time permits, we will optimize the derived closures by extensive simulations of experimental setups and practical pipelines.