在工程仿真前,进行细节移除,简化CAD模型是实际产品设计的重要步骤.此过程大大降低后续网格生成的复杂度,提升网格质量,进而减少网格生成及有限元计算失败的可能性.然而,模型简化的同时会引起仿真计算域的变化,不可避免地影响工程仿真精度,导致特征抑制误差的产生.由于缺乏相关理论支持及实际可操作方案,目前模型简化过程仍主要通过手工实现,以保证工程仿真精度,但其导致效率低下已成为CAD/CAE无缝集成的瓶颈问题.为促进问题的有效解决,文中综述聚焦于工程仿真精度可信的CAD模型简化问题,从问题的背景及意义、问题研究的整体思路,以及目前重要进展等进行介绍.问题涉及到CAD、CAE、计算物理、偏微分方程理论等多个学科或研究方向,具有鲜明的多学科交叉性.主要聚焦问题的2个核心层面:简化模型生成的几何操作和特征重要性定量评估,并对相关工作进行整体描述;最后,指出其未来的研究方向.
Simplifying an engineering design by removing geometric details before simulation can significantly reduce the complexity of downstream tasks of mesh generation and field solution computations, and improve the mesh quality and the computation reliability. However, the simplification process will inevitably change the underlying computational domain, and produce different engineering analysis results, resulting in defeaturing error. Due to the lack of theories on quantitative control of the errors induced by design simplification and associated implementation approaches, the simplification process is still mainly performed manually, very time-consuming, and remains a bottleneck issue in seamless CAD/CAE integration, and is urgently to be resolved. To help address this issue, this paper surveys related research efforts on the topic of engineering analysis reliable CAD model simplification, summarizing background and importance of the problem, an overall flowchart to resolve it, and recent research progress on resolving the important issue. This multi-discipline research topic is related to various researches on CAD, CAE, computational physics, PDE theories etc., while the survey mainly focuses on progresses on its two core issues: geometric operations to remove details or to generate the simplification model, and quantitative estimations of the impact of removing a single or a set of features to engineering analysis. Future directions on the topic are also discussed in the end.