Abstract: For the low-speed liquid injected into the high-speed strong turbulent gas flow in the same direction， the atomization is a transient-intensive spray， and there are many factors to affect and control the atomization. The distribution and characteristics of the liquid core in the air atomized flow field are analyzed. A random immersed boundary model to simulate the liquid core is developed. Combined with the large eddy simulation method， the numerical simulation of the downstream flow field of a coaxial jet air atomizing nozzle is carried out. The comparison between the simulation results and the experimental results shows that this random immersed boundary model can quickly capture the information of length and position of the liquid nucleus. When the gas-liquid momentum ratio M is 3~10000， the liquid core length can be predicted accurately. When M>10， the prediction result is much better than that of the phenomenological model. The model is capable of capturing flow field structures such as recirculation zones and large-scale vortices. The particle size of the droplets near the nozzle can be accurately predicted， especially when the gas velocity is large （bigger than 60 m/s）. The average diameter prediction error of the droplets is less than 10%.

Key words: Atomization;Flow field;Air-liquid phase interface;Numerical simulation;Large eddy simulation

摘要： 低速液体射流在高速湍流气体作用下的气液同轴射流雾化流场是瞬态密集型喷雾场，高速湍流中影响和控制雾化的因素很多。针对空气雾化流场液核分布与特征进行分析，开发了模拟液核的随机浸入体模型，结合大涡模拟方法对同轴射流空气雾化喷嘴下游流场进行数值模拟。模拟结果与实验结果对比表明：随机浸入体模型可快速捕捉液核的长度和位置信息。当气液动量比M为3~10000时，能够较准确地预测液核长度，当M>10时，其预测结果远优于唯象模型；该模型能够捕捉回流区、大尺度涡等流场结构。同时，可以准确地预测喷嘴附近的液滴粒径，特别是在气流速度较大（>60m/s）时，液滴平均直径预测误差<10%。

关键词: 雾化;流场;气液相界面;数值模拟;大涡模拟