基于格子Boltzmann异构并行计算的碎屑流非稳态动力学模型研究

碎屑流引起的群死群伤事件频发，严重危害山区经济建设。碎屑流是由大量固体颗粒相互作用而组成的复杂体系，其流变特性往往在简单剪切等稳定颗粒流动状态下提出，目前尚缺乏高效、准确的非稳态动力学模型预测碎屑流运动特征。项目采用离散单元法模拟颗粒柱坍塌运动，通过粗粒化的统计物理方法，实现离散数据（速度、受力等）向连续场的高精度转化，获取局部平均的流场信息，建立颗粒流在非稳定状态下的流变模型；采用自适应松弛时间格子Boltzmann方法，动态调节颗粒流局部流变性，使其动力学行为符合非稳态颗粒流本构模型的描述，结合流体体积法，追踪颗粒流自由表面的动态变化，对比分析离散元与格子Boltzmann颗粒柱坍塌模拟结果（外部形态及内部流场信息），验证非稳态颗粒流本构模型的准确性；采用图形处理器计算统一设备框架并行编程技术，实现格子Boltzmann数值模型并行加速计算，为碎屑流等山地灾害的防灾减灾提供技术支持。

Granular avalanches pose significant threats to human life and economic development in mountainous regions. Rheological models for granular avalanches have been proposed based on steady granular flows, such as simple shear. There is still a lack of efficient and accurate numerical models for the prediction of granular avalanche behaviors in an unsteady state. In this project, the discrete element method (DEM) is adopted to simulate granular column collapse, which is regarded as a common model case for unsteady geophysical flows. First, the discrete data from DEM simulations, including particle velocities and inter-particle forces, are converted to continuous flow fields with high accuracy using the coarse-graining technique. A rheological model for unsteady granular column collapses will be constructed based on the local averaged flow fields. Then, the lattice Boltzmann method (LBM), with a dynamic free-surface based on the volume of fluid method, will be adopted to simulate granular column collapses. In LBM, an adaptive relaxation time depending on the local shear rate will be used such that the flow behavior follows the constitutive relation for unsteady granular avalanches. By comparing the numerical results from discrete element and lattice Boltzmann simulations, in terms of the deposit morphology and the internal flow fields, the constitutive model for unsteady granular avalanches can be verified. Last but not least, the graphic processor unit will be used to accelerate the lattice Boltzmann calculations based on compute unified device architecture. The current project will establish an efficient and accurate numerical model for unsteady granular avalanches that supports heterogeneous parallel computing, which has the potential to guide the design of hazard mitigation measures.