Research On Key Technology Of The Large-scale Parallel Simulation Of Two-phase Complex Fluids

Complex fluids and its control technology play an essential role in many traditional industries and emerging industries,thus the investigation of the dynamics and the rheological characteristics of two-phase complex fluids has great scientific significance and application value.With the rapid development of high performance computing technology,numerical simulation has become an important research approach for the studying of complex fluids.However,there are still many challenges to achieve efficient large-scale parallel numerical simulation: the accurate theoretical model,a flexible parallel numerical solver,and the scalable parallel computing method.To solve these problems,this thesis mainly focus on the key technologies for large-scale parallel simulation of twophase complex fluids.The main contributions of this thesis are summarized as follows:1.Proposed two optimized theoretical models based on the molecular theory.In this thesis we firstly extended the tube-theory derived Rolie-Poly model to describing a two-phase fluid,and proposed a novel macro fluid model: the FH-RP model.Then by coupling with the BCF method,we proposed a multi-scale twofluid model for accurately calculating the viscoelastic stress through tracking massive microscopic configuration fields.The novel models will be the important basics for quantitative numerical study and could provide valuable guidance for experimental research and engineering applications.2.Designed and implemented the numerical solver of two-phase complex fluids based on parallel application framework.After analyzing the process of solving the FH-RP and the FH-BCF model,we proposed the numerical algorithms and designed the architecture of the numerical solver of two-phase complex fluids.At the last we implemented the numerical solver based on a open source CFD framework named Open FOAM,and numerical results verified the correctness and accuracy of the solution,it will be important basis for the parallel simulation of two-phase complex fluids.3.Proposed a novel nested parallel algorithm for macro-micro simulations.The time cost of macro-micro simulations is much more than pure macro simula-tions,nevertheless,the parallel scalability is limited by the grid number with traditional mesh-decomposition based parallelization approach.In order to improve the performance and reduce the simulation time of a macro-micro run,we proposed a novel two-dimensional parallel algorithm based on the mesh and Brownian configuration fields decomposition,and from the perspective of communication,the relevant optimization techniques were studied.Large-scale parallel experiments show that our method can significantly reduce the time cost of macro-micro simulations.4.Validated the techniques proposed in this paper and numerically reproduce many complex experimental observations at the first time without introducing extra parameters.We gave the detailed numerical study of the shear-induced phase separation and the non-equilibrium steady states in complex fluids,simulation results show that the FH-RP model can accurately capture the dynamic features of complex fluids under flow,even the chaotic rheological responses could be simulated without introducing any extra parameters.The numerical results reproduced many dynamic phenomena reported in literature,meanwhile the distribution of the molecule configurations in the simulations could be observed from a microscopic viewpoint.The macro-micro two-fluid model presented in the thesis provides a new approach to numerically study multi-phase viscoelastic fluids under flow.