(83c) Statistical Analysis on Large-Scale Direct Numerical Simulation of Gas-Solid Flow

Wang, L. - Presenter, Chinese Academy of sciences
Ge, W., Institute of Process Engineering, Chinese Academy of Sciences
Particle-resolved direct numerical simulation (DNS), where the computational grid for the fluid phase is reduced to below the particle size, the flow around each particle is fully resolved, and the fluid-solid interaction force is directly obtained by integrating the viscous force and pressure on the particle’s surface. It has been regarded as the most accurate numerical method for simulation of gas-solid flow. However, the main disadvantage is its huge computational cost resulting from small grid size and time step limited by Kolmogorov length and time scales. Only hundreds of particles scale is reported for DNS of gas-solid flow in the latest literature, which is very different from the number of particles in the real gas-solid flows. Therefore, small-scale simulation results are hardly to be established a true and reliable constitutive law for gas-solid flow. In order to solve the problem of computational speed and scale, we use the lattice Boltzmann method (LBM) with good parallelism to replace the traditional N-S equation to describe the gas phase, an immersed boundary method in framework of LBM has been adopt to realized the fluid-solid coupling to avoid a stair-step representation of the solid particles’ surfaces (Wang et al., 2010; Zhou et al., 2011), and the multi graphics processor units (GPUs) parallel computing is also implemented. Taking advantage of the inherent parallelism of LBM and the attractive Flops/Price ratio of GPU, we have implemented 576 GPUs parallel computing on a Mole-8.5 system and conducted the largest scale DNS of gas-solid suspensions so far, with 1,166,400 solid particles in an area of 11.5cm x 46cm for a two-dimensional system and 129,024 solid particles in a domain of 0.384cm x 1.512cm x 0.384cm for a three-dimensional system (Xiong et al., 2012). The scale of DNS data has been reached the size in traditional computational grid, which implies the really meaningful statistical results from large-scale DNS of gas-solid flows were obtained for the first time. The effects of mesoscale structure on the interaction force between gas and solid phases (Zhou et al., 2014) and the statistical properties of particles (Liu et al., 2017) were explored, which may provide the corresponding constitutive relation and detailed microscopic information for discrete particle simulation and two-fluid model.


  1. Limin Wang, Guofeng Zhou, Xiaowei Wang, Qingang Xiong, Wei Ge. 2010. Direct numerical simulation of particle-fluid systems by combining time-driven hard-sphere model and lattice Boltzmann method. Particuology, 8: 379-382.
  2. Guofeng Zhou, Limin Wang, Xiaowei Wang, Wei Ge. 2011. Galilean-invariant algorithm coupling immersed moving boundary conditions and Lees-Edwards boundary conditions. Physical Review E 84, 066701.
  3. Qingang Xiong, Bo Li, Guofeng Zhou, Xiaojian Fang, Ji Xu, Junwu Wang, Xianfeng He, Xiaowei Wang, Limin Wang, Wei Ge, Jinghai Li. 2012. Large-scale DNS of gas–solid flows on Mole-8.5. Chemical Engineering Science 71: 422–430.
  4. Guofeng Zhou, Qingang Xiong, Limin Wang, Xiaowei Wang, Xinxin Ren, Wei Ge. 2014. Structure-dependent drag in gas-solid flows studied with direct numerical simulation. Chemical Engineering Science, 116: 9-22.
  5. Xiaowen Liu, Limin Wang, Wei Ge. 2017. Meso–scale statistical properties of gas–solid flow—a direct numerical simulation (DNS) study. AIChE Journal 63:3-14.