Computational & Technology Resources
an online resource for computational,
engineering & technology publications
PROCEEDINGS OF THE SECOND INTERNATIONAL CONFERENCE ON PARALLEL, DISTRIBUTED, GRID AND CLOUD COMPUTING FOR ENGINEERING
Edited by: P. Iványi and B.H.V. Topping
Parallel Algorithms for Particle-Turbulence Two-Way Interactions in a Weakly Compressible Jet by Direct Numerical Simulation
D.B. Li, J.R. Fan, F.X. Yi, S.Q. Lu and K.F. Cen
State Key Laboratory of Clean Energy Utilization, Zhejiang University, Hangzhou, P.R. China
D.B. Li, J.R. Fan, F.X. Yi, S.Q. Lu, K.F. Cen, "Parallel Algorithms for Particle-Turbulence Two-Way Interactions in a Weakly Compressible Jet by Direct Numerical Simulation", in P. Iványi, B.H.V. Topping, (Editors), "Proceedings of the Second International Conference on Parallel, Distributed, Grid and Cloud Computing for Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 12, 2011. doi:10.4203/ccp.95.12
Keywords: SPMD, particle-laden jet, explicit finite different method, parallel computing, two-way, domain decomposition.
SPMD parallel computation of a weakly compressible particle-laden jet flow is presented in our paper. The parallel computation is conducted with an explicit finite difference method using the direct numerical method on an IBM Linux Cluster. The conservation equations, boundary conditions, domain decomposition method, and the solving processes are carefully presented in order to give a clear understanding of the parallel computing of a weakly compressible particle-laden turbulent flow using the explicit finite difference method, which is scarce in the literature [1,2,3]. In addition, a large number of particles are tracked in a Lagrangian frame. Point forces are used to represent the back effect of the particles on the turbulence. Two-way coupled forces are used to consider the effect of the interaction between particles and fluid. The speedup factor and parallel computational efficiency are presented with different domain decomposition methods (one domain decomposition A or two domain decomposition B).
A one domain decomposition A or a two domain decomposition B for the parallel computing is proposed in this paper together with the corresponding particle parallel algorithm . From our extensive numerical research, we find that the two domain decomposition B is more parallelly efficient for particle-laden jet flow simulation than that of one domain decomposition A based on the analysis of speedup factor and efficiency, but the two domain decomposition B is more complex than that of the one domain decomposition A considering the message exchange between particles and fluid.
In order to validate our two-phase code, the numerical simulation results are compared carefully with experimental data, considering the mean velocity, the Reynolds stress for the fluid and particles [1,5,6,7,8]. To some extent, our DNS code is useful for large-scale particle-laden turbulent flow.
purchase the full-text of this paper (price £20)