TL;DR
This paper introduces a novel hierarchical semantic mid-surface abstraction method for thin-walled models, crucial for efficient engineering analysis. By decomposing models based on rib features and incorporating semantic information, the method achieves improved accuracy, topology, and reusability compared to existing techniques.
What Problem Does This Paper Address?
Computer-Aided Design (CAD) and Computer-Aided Engineering (CAE) integration often requires simplifying complex 3D models, especially thin-walled structures common in industries like automotive and aerospace. Mid-surface abstraction is a key simplification technique that reduces a thin-walled solid model to a 2D surface, significantly speeding up Finite Element Analysis (FEA). However, existing methods struggle with complex geometries, freeform surfaces, and maintaining accurate topology. Furthermore, these methods often lack reusability, meaning the entire abstraction process must be repeated even for minor design changes. This paper addresses these limitations by proposing a method that leverages hierarchical semantic structures and rib decomposition for more robust and efficient mid-surface generation and reuse.
Methods
The proposed method introduces a hierarchical semantic structure to represent thin-walled models. This structure captures both connection semantics (how parts connect) and hierarchical semantics (the importance or dependency of parts). The core methodology involves five steps:
- Hierarchical Semantic Structure Definition: A new structure is defined for both the thin-walled and mid-surface models.
- Rib Feature Recognition and Organization: Rib features, which enhance structural stiffness, are identified and organized within the hierarchical structure. This involves a novel, efficient graph-based approach using Face Groups (FGs) and Face Group Pairs (FGPs).
- Model Decomposition: The model is decomposed into sub-regions based on the identified rib features and the hierarchical structure, using an Euler operation-based method.
- Mid-Surface Abstraction: An adaptive abstraction method is employed, using offset operations for simpler rib features and discretization (specifically, a Delaunay mesh method with a hybrid point refinement) for complex main elementary sub-regions.
- Mid-Surface Reuse: The generated mid-surface model, enriched with hierarchical semantic information, can be efficiently updated when local modifications occur to the original design, avoiding complete regeneration.
Datasets
Not discussed in the paper. The paper focuses on a proposed method and demonstrates its performance using example models, including artificial thin-walled models with simple and complex ribs, and a model with 'n x n' pocket patterns.
Key Insights and Ideas
- Hierarchical Semantics are Key: The paper argues that incorporating hierarchical semantic information (beyond just geometric connectivity) is crucial for robust mid-surface abstraction and, importantly, for enabling model reuse.
- Rib Decomposition as a Foundation: Focusing on rib features as the primary decomposition element is an effective strategy for handling the complexity of thin-walled structures.
- Adaptive Abstraction: Using different abstraction techniques (offset for ribs, discretization for main parts) based on geometric complexity is more efficient and accurate than a one-size-fits-all approach.
- Efficient Rib Recognition: The proposed FG/FGP-based graph approach significantly improves the efficiency of rib recognition compared to traditional methods.
- Reusability through Semantics: The hierarchical semantic structure directly enables efficient regeneration of the mid-surface after local design changes, a major advantage for product development workflows.
Key Results
- Improved Accuracy: The proposed method demonstrates significantly better geometric accuracy, with R-errors generally less than 1‰, outperforming CAT and projection methods which show errors up to 10% and 1% respectively.
- Enhanced Topology: The method aims to produce mid-surface models with better topological integrity, particularly by addressing connection semantics.
- Increased Efficiency: The rib recognition method is shown to be more efficient, with computation time increasing less dramatically with model complexity compared to a typical method.
- Effective Reuse: The paper demonstrates the ability to efficiently regenerate the mid-surface model after local modifications, as shown in case studies.
Limitations and Future Work
- Non-Thin-Walled Areas: The authors acknowledge that the current approach is primarily designed for thin-walled models and may not handle models with significant non-thin-walled areas.
- Future Research: The paper suggests that future research will focus on extending the hierarchical semantic structure to better handle models with mixed thin-walled and non-thin-walled regions.
Summary updated