Study on the Subgrid Scale Modeling for Particle-Laden Turbulent Flows

本文对流固两相湍流进行了直接数值模拟（DNS），以考虑颗粒负载湍流大涡模拟（LES）中的亚网格尺度（SGS）建模。首先，我们开发了一个有限差分格式来解决流体湍流和粒子运动之间的相互作用。该方法直接模拟了长度尺度大于计算网格间距数倍的固体颗粒周围的流动。粒子的运动受表面压力和粘性应力的积分以及物体力（如重力）的控制。该方案不包括流体湍流、粒子运动或它们之间相互作用的模型。然后，我们通过将其应用于固定在均匀流中的球体的三维流来检验我们的新方法。在阻力系数和非定常涡脱落方面，数值计算结果与实验结果吻合较好。当涡旋脱落发生时，尾迹诱导的湍流能量在粒子附近并没有完全消散。这种非平衡性质在粒子的阻力改变后得到增强。其次，我们模拟了一个强迫各向同性湍流和湍流在一个平面通道包括球粒子。粒子形成了更大的结构，如簇，从而增加了流体湍流的能量来源。在另一种装置中，固定粒子的相对位置，观察没有粒子团簇的湍流调制。在这种情况下，由于能谱向高波数范围的移动，总能量有时会减少，即使粒子诱导额外。在壁面湍流中，颗粒在通道内分布不均匀。在这种情况下，由于速度梯度，脱落涡被拉伸到顺流方向，从而产生雷诺剪应力。这些带有颗粒的湍流的DNS结果表明，在SGS建模中需要考虑一些重要的特征。例如，非定常旋涡脱落引起的非平衡性质、流体湍流能量注入的多尺度、颗粒分布对湍流长度尺度的调制以及由于脱落旋涡拉伸而产生的雷诺应力。少

We carried out a direct numerical simulation (DNS) of fluid-solid two-phase turbulent flows to consider the subgrid scale (SGS) modeling in the large-eddy simulation (LES) of particle-laden turbulence.First, we developed a finite-difference scheme to resolve the interaction between fluid turbulence and particle motion.In our method, the flow around a solid particle having a length scale several times larger than the spacing of computational grid, is directly simulated.The particle motion is governed by the surface integral of pressure and viscous stress together with the body force such as gravity.This scheme includes no models in fluid turbulence, particle motion or interaction between them.Then, we examined our new method by applying it to the three-dimensional flow past a sphere fixed in a uniform stream.As for the drag coefficient as well as the unsteady vortex shedding, numerical results agreed with experimental results.When the vortex shedding took place, turbulence energy induce … More d through wake was not fully dissipated in the region near a particle.This non-equilibrium nature was enhanced just after the change in drag force of a particle.Next, we simulated a forced isotropic turbulence and turbulent flow in a plane channel including sphere particles.Particles formed larger structure such as clusters, resulting in an additional scale of energy source for fluid turbulence.In an another setup, the relative position of particles was fixed to observe the turbulence modulation without particle clusters.In this case, total energy was sometimes reduced due to the shift of energy spectra into high wavenumber range, even though particles induced extra.In the wall turbulence, particles were distributed non-uniformly in the channel.In this case, the shed vortices was stretched into the streamwise direction due to the velocity gradient, resulted in the production of Reynolds shear stress.These DNS results of particle-laden turbulent flows suggested some important characteristics to be accounted for in SGS modeling.They are non-equilibrium nature due to unsteady vortex shedding, multiple scale of energy injection to fluid turbulence, the modulation in turbulence length scale by particle distribution, and the Reynolds stress production due to the stretch of shed vortices, for example. Less