Numerical Analysis of Meso-Scale Particle Structure formed in Gas-Solid Turbulence

1. 垂直通道内气固向下流动的LBS对垂直通道内载颗粒向下的湍流流动进行了双向耦合的数值模拟，其条件与斯坦福大学研究小组的条件相对应。用与单相流相同的模型用LES模拟流体运动，用拉格朗日方法模拟单个粒子的运动。考虑了粒子间的碰撞。首先，通过对气体流场进行滤波，研究了LES网格分辨率对颗粒扩散的影响。网格分辨率的影响与粒子弛豫时间与流体涡旋与粒子相互作用时间之比所定义的斯托克斯数相关。提出了粒子运动的LES网格分辨率判据。随后，对实验中使用的三种颗粒进行气固流动的LES。双向耦合的模拟结果表明，粒子在近壁区有向高速条纹集中的趋势，而单向耦合的模拟结果则相反。通道中心区域粒子云的预测模式与实验结果吻合较好。本文还对二维循环流化床中的颗粒团簇结构进行了数值模拟。考虑通道壁的模拟比不考虑通道壁影响的模拟预测出更小的分级尺寸。研究发现，粒径分布对团簇规模的影响较小。

1. LBS of downward gas-solid flows in a vertical channelNumerical smuations with two-way couping were performed for downward particle laden tutbulent flows in a verticul channel, of which conditions corrspond to those by a group of Stanford University.Fluid motion was culculated by using LES with the same models as in single phase flows and the motin of individual particles by the Lagrangian method.Particle-particle collisin was taken into considertion.First, the effect of grid resolution of LES on the particle diffusion was investigated by appling filtering to gas-flow fields.The effect of grid resolution was correlated with Stokes number defined by the ratio of particle's relaxation time to the interaction time between a fluid eddy and a particle.An criterion of grid resolution of LES for particle motion was proposed.Subsequently, LES of gas-solid flows was performed for three kinds of particles used in the corresponding experiment.The present simulations with two-way coupling predicted that particles tend to concentrate to high-speed streaks in the near-wall region, while the previous ones with one-way coupling predicted the contrary tendency.The predicted patterns of particle clauds in the channel-center region agree well with the experiment.2. Structure of paticle clusterThe structure of particle cluster in a two-dimensional circulation fluidized bed is also simulated numerically.The simulation taking account of the chanel walls predicted smaller claster size than that without the effect of the chanel walls.It was found that the effect of particle size distribution on the scale of clusters is small.