Representing large-scale motions and topological changes in the finite volume (FV) framework, while at the same time preserving the accuracy of the numerical solution, is difficult. In this paper, we present a robust, highly efficient method designed to achieve this capability. The proposed approach conceptually shares many of the characteristics of the cut-cell interface tracking method, but without the need for complex cell splitting/merging operations. The heart of the new technique is to align existing mesh facets with the geometry to be represented. We then modify the matrix contributions from these facets such that they are represented in an identical fashion to traditional boundary conditions. The collection of such faces is named a Generalised Internal Boundary (GIB). In order to introduce motion into the system, we rely on the classical ALE (Arbitrary Lagrangian-Eulerian) approach, but with the caveat that the non-time-dependent motion of elements instantaneously crossing the interface is handled separately from the time dependent component. The new methodology is validated through comparison with: a) a body-fitted grid simulation of an oscillating two dimensional cylinder and b) experimental results of a butterfly valve.
翻译:代表数量有限(FV)框架的大规模动议和地形变化,同时保持数字解决办法的准确性,是困难的。本文提出一种为实现这一能力而设计的有力、高效的方法。提议的方法在概念上分享切入细胞界面跟踪方法的许多特点,但不需要复杂的细胞分裂/合并操作。新技术的核心是使现有的网格方面与要代表的几何结构相协调。然后我们修改这些方面的矩阵贡献,使其与传统边界条件相同。这些面孔的收集称为通用的内部边界(GIB)。为了将运动引入系统,我们依靠传统的ALE(任意拉格朗吉亚-尤莱里安)方法,但告诫说,不依赖时间的元素瞬间穿越界面的动作与时间依赖部分分开处理。新的方法通过与以下的比较得到验证:a)为结构设计的电网模模拟,显示2维的气瓶和b)一个阀门的试验结果。