EI、Scopus 收录
中文核心期刊

基于格子玻尔兹曼方法的局部网格加密算法——粗细网格间的数据转换

LOCAL GRID REFINEMENT APPROACH FOR LATTICE BOLTZMANN METHOD: DISTRIBUTION FUNCTION CONVERSION BETWEEN COARSE AND FINE GRIDS

  • 摘要: 格子Boltzmann方法作为一种高效的介观计算流体力学方法在过去20多年里得到快速发展, 其相对较高的计算效率和灵活性使其可以适用于各种复杂流动的模拟. 然而标准的格子Boltzmann方法只能使用均匀的直角网格, 这种网格排布方式并不利于复杂流动的计算. 为此, 基于格子Boltzmann方法的局部网格加密算法在文献中被提出. 该算法需要在局部加密的界面处将粗细网格间的分布函数转换后交换. 目前分布函数的转换方式大多是在没有源项的情况下推导的, 而且现存考虑源项时转换公式的推导也都是基于Chapman-Enskog展开; 其推导过程相对复杂, 且需要对分布函数的非平衡态部分做一阶Chapman-Enskog近似, 这有可能会限制局部网格加密算法在高阶格子Boltzmann方法中的应用. 文章在忽略时空离散误差的前提下, 以保证连续分布函数变量以及物理松弛系数一致为基础, 构建了一套规范且简洁的粗细网格间在考虑任意源项时, 分布函数转换关系的推导过程, 该方法不依赖于Chapman-Enskog展开以及Chapman-Enskog近似, 且该方法既可以适用于单松弛碰撞模型也可以适用于多松弛碰撞模型. 此外, 还从理论上证明了, 保证粗细网格间非平衡态部分的一阶 Chapman-Enskog 近似一致, 便可以保证整个非平衡态部分的一致, 这将有助于扩展局部网格加密算法中转换关系的应用范围. 最后, 通过对强迫泰勒−格林涡流动、平板泊肃叶流中对流−扩散问题和顶盖驱动方腔流动进行数值模拟, 良好的数值结果证实了转换关系对复杂源项的适应性以及局部网格加密技术在处理复杂流动问题方面的优势. 同时, 通过对一维剪切波问题的模拟, 发现由局部网格加密引起的数值黏性与加密区域的选取有很大的关系.

     

    Abstract: Lattice Boltzmann method, as an efficient mesoscopic computational fluid dynamics method, has been developed rapidly in the past two decades. Its relatively high computational efficiency and flexibility make it suitable for the simulation of various complex flows. However, due to its own limitations, the standard lattice Boltzmann method typically utilize uniform rectangular grid, which is not suitable for the simulation of complex flows. Therefore, local grid refinement based on the lattice Boltzmann method has been considered by many researchers. For this purpose, the distribution functions between coarse and fine grids need to be converted at the interface of coarse and fine grids. At present, most conversion methods of distribution function are derived without the presence of the source term, and the previously limited derivation of conversion formulas considering the source term was based on the first-order Chapman-Enskog expansion, which is relatively complex and may limit the application of local grid refinement algorithm in higher-order lattice Boltzmann methods. In this paper, we provide a concise derivation to relate the distribution functions between coarse and fine grids considering an arbitrary source term, based on consistency requirements of the distribution function of the continuous Boltzmann equation between coarse and fine grids. The proposed method is independent of the Chapman-Enskog expansion and Chapman-Enskog approximation, and can be applied to both single relaxation time and multiple relaxation time collision models. In addition, this paper also proves theoretically that the consistency of the first-order Chapman-Enskog approximation of the non-equilibrium distribution between the coarse and fine girds can ensure the consistency of the entire non-equilibrium distribution, which expands the applicability of the previous conversion relationship. Finally, these theoretical results are validated by numerical simulations of a forced Taylor-Green vortex flow, convection-diffusion in a planar Poiseuille flow and lid-driven cavity flow. The good numerical results confirm the adaptability of the conversion relation in the presence of complex source terms and the advantages of local grid refinement technology in dealing with complex flow problems. At the same time, through the simulation of one-dimensional shear wave problem, it is found that the numerical viscosity caused by local grid refinement has a great relationship with the selection of refinement region.

     

/

返回文章
返回