426459 Euler-Euler Anisotropic Gaussian Mesoscale Direct Numerical Simulation of Cluster-Induced Turbulent Flow

Monday, November 9, 2015: 2:05 PM
254C (Salt Palace Convention Center)
Bo Kong1,2, Heng Feng3, Jesse S. Capecelatro4, Olivier Desjardins5 and Rodney O. Fox1,6, (1)Applied Mathematics and Computational Sciences, Ames Laboratory (U.S. DOE), Ames, IA, (2)Department of Chemical and Biological Engineering, Iowa State University, Ames, IA, (3)Department of Thermal Engineering, Tsinghua University, Beijing, China, (4)The Center for Exascale Simulation of Plasma-coupled Combustion, University of Illinois at Urbana-Champaign, Urbana, IL, (5)Sibley School of Mechanical and Aerospace Engineering, Cornell University, Ithaca, NY, (6)Chemical and Biological Engineering, Iowa State University, Ames, IA

Gas-particle flows are common in many fields of engineering, such as in fluidized beds and risers, which are widely used in a variety of chemical processes. The accurate simulation of such flows is crucial for the design and optimization of their industrial applications. Although gas-particles flows in industrial applications are often turbulent, available multiphase turbulence models in commercial CFD codes often lack a rigorous conceptual foundation. In our previous works, the exact Reynolds-averaged (RA) equations were derived for the particle phase in a collisional gas-particle flow (Fox, 2014), and detailed Euler-Lagrange particle simulations of cluster-induced turbulence (CIT) were performed to aid the development of this model (Capecelatro et.al. 2013, 2014). However, sophisticated filtering techniques have to be used to extract particle-phase information consistent with the Eulerian turbulence model from Euler-Lagrange simulations, and the results can be sensitive to various aspects of the filtering process. By comparison, Euler-Euler approaches of gas-particle flows can directly provide particle-phase turbulence statistics, and are well suited for high performance computations. In the current work, a novel Euler-Euler Anisotropic Gaussian approach (EE-AG), in which particle velocities are assumed to follow a multi-variate anisotropic Gaussian distribution, is used to perform mesoscale DNS of CIT cases, which were studied in our previous Euler-Lagrange simulations. A constant Stokes drag model is used to compute the momentum exchange between phases. A three-dimension Hermite Quadrature formulation is used to calculate kinetic flux for ten velocity moments in a finite-volume frame work. The Bhatnagar-Gross-Krook (BGK) model is applied here to account for the inelastic particle collisions. Finally, the particle-phase volume-fraction and momentum equations are coupled with the Eulerian solver for the gas phase. This approach is implemented in an open-source CFD package, and detailed simulation results are compared with Euler-Lagrange simulations. The results demonstrat that the AG assumption for particle velocity is valid and this novel method can be used to perform mesoscale DNS for gas-particle flows. The advantage of EE-AG approach over traditional kinetic theory model with two-fluid method approach (KT-TFM) in gas-particle flow simulation will also be discussed. 

Extended Abstract: File Not Uploaded
See more of this Session: Fundamentals of Fluidization II
See more of this Group/Topical: Particle Technology Forum