Abstract The computational fluid dynamics (CFD) simulators emerged as an excellent tool to support engineering problem-solving in the past decades. There are limits to the application of a CFD simulator, for example, the validation need of the model, and the computation demand of the simulation. With multiple core computers, the computation time can be lowered, and GPU computing can also be a way to decrease the computation demand. But in some cases, for example in catalyst beds, the number of individual particles is too high to calculate with a CFD simulator. In this study, we show a way to segregate the geometry of the device into smaller parts (decomposition of the geometry) and calculate only the parts of the simulation at a time, instead of whole. In this way, the computation need can be significantly lowered, without losing the crucial information, which can be stored between the geometrical steps. In this article, a framework was developed for the segregation of the geometry. The operation of the framework is shown using different case studies.

中文翻译：

---

用于复杂几何计算的几何分离 CFD 模型求解框架

摘要 在过去的几十年里，计算流体动力学 (CFD) 模拟器成为支持工程问题解决的优秀工具。CFD 模拟器的应用存在限制，例如模型的验证需求和模拟的计算需求。使用多核计算机，可以降低计算时间，GPU 计算也是降低计算需求的一种方式。但在某些情况下，例如在催化剂床中，单个颗粒的数量太多，无法使用 CFD 模拟器进行计算。在这项研究中，我们展示了一种将设备的几何结构分成更小的部分（几何结构的分解）的方法，并且一次只计算模拟的部分，而不是整个部分。通过这种方式，可以显着降低计算需求，不会丢失关键信息，这些信息可以存储在几何步骤之间。在本文中，开发了一个用于几何分离的框架。使用不同的案例研究展示了该框架的操作。