基于数据驱动的气液旋流分离群体平衡模型构建及机理研究

Gas-liquid cyclone separation is a common two-phase flow process in various industries, such as the petrochemical industry, energy and power industry, and nuclear industry. Due to the droplet coalescence and breakup phenomena as well as droplet deposition and entrainment phenomena, the physical essence of gas-liquid cyclone separation under high gas-liquid ratio is a complex multi-dispersed and multi-phase system with changing droplet size distribution. The accurate prediction for droplet size distribution and separation mechanism considering the evolution of droplet size distribution are two key issues for the design and optimization of gas-liquid cyclone separator. In order to overcome these two problem, this project will carry out the following researches. Firstly, the droplet coalescence and breakup models induced by three-dimensional strong swirling turbulence are established in the framework of entire energy spectrum, and the deposition and entrainment models are proposed by applying an inverse problem technique. Then, based on the data-driven physics-evaluated machine learning framework, the closure relations that can be applied to different structures and system conditions are obtained. Based on these two studies, a novel PBM with high accuracy and strong adaptability is constructed and used to reveal the mechanism of gas-liquid cyclone separation considering the evolution of droplet size distribution. With these efforts, it is expected to provide scientific guidance and technical support for the design and optimization of gas-liquid cyclone separator.

气液旋流分离是石油化工、能源动力和核电等工业领域中常见的两相流过程，在高气液比下其物理本质为液滴聚并破碎和沉积夹带作用下液滴粒径分布不断变化的复杂多分散多相体系。液滴粒径分布的准确预测以及考虑液滴粒径分布演化的机理性认识是气液旋流分离器研发的两大难题。为了克服这两大难题，本项目拟开展如下研究：在整个湍流能谱范围内建立三维各向异性强旋湍流下的液滴聚并和破碎模型，通过反问题方法建立液滴沉积和夹带模型，实现液滴聚并破碎和沉积夹带现象的精细建模；采用数据驱动的“物理评估机器学习框架”，构建适用于不同旋流器结构和系统条件的群体平衡模型封闭关系；通过上述研究，建立高精度、强适应性的气液旋流分离群体平衡模型，实现液滴粒径分布的准确预测，并采用该模型揭示考虑液滴粒径分布演化的气液旋流分离机理。研究成果可为高效紧凑型气液旋流分离器的研发提供理论指导和技术支撑。

气液旋流分离是石油化工、能源动力和核电等工业领域中常见的两相流过程，在高气液比下其物理本质为液滴聚并破碎和沉积夹带作用下液滴粒径分布不断变化的复杂多分散多相体系。液滴粒径分布的准确预测以及考虑液滴粒径分布演化的机理性认识是气液旋流分离器研发的两大难题。为了克服这两大难题，本项目以管柱式气液旋流分离器为对象，建立了考虑液滴粒径分布演化的气液旋流分离群体平衡模型，并与欧拉-欧拉模型和欧拉液膜模型进行耦合，实现了气液旋流分离过程的准确预测；基于开源计算流体力学软件OpenFOAM和IDEAL算法，开发了适用于任意多面体网格的高效健壮单相流、两相流气液旋流分离器数值计算方法，实现了气液旋流分离的高效数值计算；基于群体平衡模型，研究并揭示考虑液滴粒径分布演化的气液旋流分离机理。研究成果可为高效紧凑型气液旋流分离器的研发提供理论指导和技术支撑。基于以上研究，发表SCI论文6篇、EI论文1篇，协助培养研究生2名。