粘弹性非混相多相流体的多尺度建模与数值方法研究

粘弹性非混相多相流广泛存在于自然界和工业界。它不仅具有单相粘弹性流固有的记忆效应，而且其相界面两侧流体的性质差异还会引起复杂的界面演化，表现出丰富的物理现象和流体动力学机制。目前，该流体的数值研究面临两个难题：传统宏观模型建模质量难以满足需求和高威森伯格数问题模拟困难。由于该流体在介观尺度具有结构化组织和时间动态特征，多尺度模型拥有着宏观模型所没有的优秀建模质量和模型扩展能力。基于此，本项目拟耦合现有确定性多尺度方法和多相格子玻尔兹曼方法（LBM），发展出多尺度Lattice-Boltzmann-Fokker-Planck模型。为克服求解高维森伯格数问题时遇到的数值耗散高、稳定性差等问题，本项目拟基于CPU-GPU异构平台，引入对于对流项具有零耗散特点的LBM和高稳定性的递归正则化技术，为相关复杂物理问题研究和工业数值计算设计出低耗散、稳健、高效的算法，并通过相关问题的计算来验证其准确性。

Viscoelastic immiscible multiphase flow exists widely in nature and industry. It has the memory effect inherent to single-phase viscoelastic fluids, and the difference in the properties of the fluids across the phase interface will cause complex interfacial evolution, which make it exhibit rich physical phenomena and hydrodynamic mechanisms. Currently, the numerical study of this fluid faces two challenges: the difficulty to meet the demand of quality by traditional macroscopic modeling and the difficulty of simulating of high Weissenberg number problem. Due to the structured organization and the temporal dynamics of viscoelastic immiscible multiphase fluids at mesoscopic scale, mesoscopic scale models possess excellent modeling quality and model scalability that macroscopic scale models do not. Based on this, this project will develop a multiscale Lattice-Boltzmann-Fokker-Planck model by coupling the existing deterministic multiscale method and the multiphase lattice Boltzmann method (LBM). In order to overcome high numerical dissipation and poor stability in solving the high Weissenberg number problem, this project proposes to introduce LBM with zero dissipation for convective terms and highly stable recursive regularization technique to design a low-dissipative, robust and efficient algorithm based on CPU-GPU heterogeneous platform for the research into related complex physical problems and the numerical computation of industrial problems, and verify its accuracy by computing the relevant problems.