Research On Key Technology Of Macro-micro Coupling Parallel Simulation Of Viscoelastic Fluids

Control techniques for viscoelastic fluids and viscoelastic fluids are widely used in industrial production.In recent years,high-performance computing has developed rapidly.And research on large-scale parallel numerical simulation of viscoelastic fluids can provide an effective scientific and practical reference for fluid engineering calculations.At present,the mainstream numerical solvers for viscoelastic fluids still have obvious defects in parallel scalability,and cannot effectively utilize large-scale parallel computing resources.Therefore,concentrating on the challenges of large-scale parallel computing of viscoelastic fluids,this thesis developed research on the design of macro-micro numerical solver,code implementation and parallel algorithm optimization.The main work and innovation are mainly reflected in the following points:1.The design and implementation of macro-micro multiscale parallel numerical solver,BCF-FOAM.This thesis developed a multi-scale numerical solver BCF-FOAM for viscoelastic fluids on Open FOAM,an open source CFD software platform.Through the in-depth analysis of the basic principle and mathematical model of the Brown Configuration Field(BCF)method,the overall architecture of the numerical solver BCF-FOAM is designed on Open FOAM.According to the solution process of macro-micro numerical simulation of viscoelastic fluids,we designed and implemented a macro-micro numerical algorithm coupling with BCF method based on PISO iterative algorithm provided by Open FOAM.And then,we tested and analyzed whether BCF-FOAM can be used to simulate the viscoelastic fluids within the tolerances through a two-dimensional simulation application for the dilute polymer solution.2.The implementation of hybrid decomposition parallel algorithm for simultaneously decomposing meshes and configuration fields based on Open FOAM.Through the in-depth analysis of the computational characteristics of the macro-micro parallel numerical simulation of viscoelastic fluids,we found that the microscopic equations of the configuration field in BCF can be solved independently.Based on the Open FOAM platform,this thesis implements a parallel numerical algorithm for the hybrid decomposition of mesh and configuration field.Based on the mesh region decomposition in the traditional CFD parallel computing scheme,this algorithm introduces the decomposition of configuration field,which significantly improves the parallel scale of the simulation problem.3.The multi-threaded optimization and communication optimization of hybrid decomposition algorithm.In order to give full play to the advantages of multi-core processors and explore the potential data parallelism of programs,we introduced Open MP's multi-threaded technology to optimize the macro-micro numerical solver using MPI and Open MP multi-level parallel framework.It provides a way for us to work on multi-threaded optimization for the configuration field in the next step.By deeply analyzing the communication process and the average communication time in the hybrid decomposition algorithm,the broadcast and reduction operations in the group communication are used to replace the point-to-point communication operation between the original master process and the corresponding slave processes,and the execution time of the optimized numerical solver is reduced by nearly 20%.