Research On Complex Fluid Simulation And Parallel Optimization With OpenFOAM

The continuous improvement of High Performance Computing Technology has led to the rapid development of the Large-scale Science and Engineering computing represented by Computational Fluid Dynamics.As one of the research hotspots in Computational Fluid Dynamics,numerical simulation of complex fluids has a great scientific significance and application value.At present,complex fluids simulation technology can not meet the actual needs in solver designing,simulation accuracy and computational efficiency,which needs to be optimized in terms of theoretical model and parallel optimization.Based on an open-source CFD framework named Open FOAM,this thesis has studied the theoretical model,solver designing and parallel optimization of two-phase complex fluids to solve these problems.The main contributions of this thesis are shown as follows:1.Designing and implementing the numerical solver of two-phase complex fluids:FHRP-FOAM.Using the two-fluid model as the main framework,this thesis couples the FloryHuggins mixing free energy function and the Rolie-Poly constitutive model based on the molecular theory to build a two-phase complex fluids model.Based on the theoretical model and parallel CFD framework named Open FOAM,this thesis has designed and implemented a numerical solver for two-phase complex fluids named FHRP-FOAM,which is applicable for large-scale parallel numerical simulations of two-phase complex fluids.2.Studying the non-equilibrium steady states in phase separation of binary sheared entangled polymer mixtures with the FHRP-FOAM solver,and confirming that the existence of non-equilibrium steady state depends on the magnitude of viscoelastic contribution in the system.With the help of FHRP-FOAM solver,this thesis has studied the non-equilibrium steady states in phase separation of binary sheared entangled polymer mixtures.This thesis has a quantitative analysis on the effect of viscoelasticity in the fluid and confirms the irreplaceable role of viscoelasticity in forming the non-equilibrium steady states,which extended conclusions of previous studies.3.Analyzing the bottleneck of parallel behavior for FHRP-FOAM solver.Proposing a parallel optimization method based on the asynchronous global reduction and achieving a good parallel behavior.This thesis has tested the strong scaling and weak scaling of the solver,analyzed the bottleneck of parallel behavior for FHRP-FOAM solver,and proposed an optimization method based on the asynchronous global reduction.This method replaces the blocking global reduction with the nonblocking global reduction,and combines the loop unrolling with communication overlap to reduce the communication cost.Experimental results show that this method could dramatically decrease the execution time by 10% ? 30% and achieve a good optimizational behavior.