網格在有限元素分析扮演極為重要的角色，網格會直接影響模擬分析的準確度與效率，而網格建立卻也是最耗時的工程，以射出成型中最常見的薄殼元件為例，薄殼元件上有許多的凸起特徵，這些特徵包含肋、管、柱或是一些複雜支撐結構或造型等，若使用Tetrahadron自動建立網格，這些特徵容易造成網格品質下降，而最常見的處理方式是將這些特徵分解，並依照不同形狀特徵，個別進行網格化，以改善網格品質，而以往這些工程都由工程師手動執行，而花費大量時間，若能將此步驟自動化，更能加快網格建構。而本研究是針對薄殼元件，將其內部之凸起物特徵自動辨識並分解，在辨識過程中會記錄凸起物彼此間的關聯性及面的配對性，以便作為後續分解以及如何建立網格的參考，本研究最後也會執行簡易的自動拆解，驗證拆解完後的模型網格品質改善。;Solid meshes play an important role in finite element analysis. They affect the accuracy and efficiency of the numerical analysis and simulation. They also affect the computational time significantly. Thin shell parts commonly appear in injection molding and contain many extrusion features, such as ribs, tubes, complex supporting structure and molding. If tetrahedron meshes are employed for these features, it usually results in poor quality mesh. A common solution for this issue is decomposing these features and generating appropriate types of meshes for each of them. Conventionally, this process is done manually, which requires substantial operating time. If extrusion features on CAD models can be recognized and decomposed automatically, it could speed up the process of mesh generation. The aim of this research is to develop a feature recognition algorithm for internal extrusion features on thin shell parts, and to develop a feature decomposition algorithm to separate them from the model. In the proposed feature recognition algorithm, the inner and outer faces of a model are separated first. The relationship between extrusions is then evaluated, and the paired faces on each extrusion are determined. The recorded information is then employed for the decomposition of extrusions. An automatic decomposition algorithm was developed for simple CAD models. The quality of the meshes generated was also verified.
