CAD Face Abstraction for High-Quality Feature-Based Meshing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing CAD systems face challenges in generating high-quality finite element meshes due to small geometric edges and faces, particularly in complex parts like automotive carbody panels and aerospace structures, which can lead to poor mesh quality and failure to meet industry-specific criteria.

Innovation Solution

A method for CAD operations that classifies faces by curvature, identifies and merges sliver and narrow blend faces, restores original faces in high-curvature zones, and processes shared edges to produce merged faces, while preserving geometric continuity, resulting in a high-quality mesh that captures essential features.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If traditional mesh generation methods are used for complex anatomical structures, then mesh quality may be compromised, but the process becomes time-consuming and requires manual intervention

Engineering Contradiction:
Improvemesh qualityVSAvoidmesh generation time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The system automatically generates high-quality meshes by having the computational algorithm self-correct and self-optimize the mesh generation process without human intervention. The method uses automated quality assessment and iterative improvement to maintain high mesh quality while eliminating manual time investment.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system dynamically adjusts mesh generation parameters based on the complexity of the anatomical structure being processed. By changing parameters such as element density, refinement levels, and quality thresholds adaptively, the system maintains high mesh quality for complex structures while reducing processing time for simpler cases.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If detailed anatomical models are created with high mesh density, then measurement precision improves, but computational resources and processing time increase

Engineering Contradiction:
Improveanatomical measurement accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system applies different mesh densities to different regions of the anatomical model based on their importance and geometric complexity. Critical regions requiring high measurement precision receive dense meshing, while less critical areas use coarser meshing, thereby maintaining accuracy where needed while reducing overall computational complexity.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The anatomical model is divided into multiple segments or regions with different mesh density requirements. This segmentation allows the system to apply computational resources efficiently, creating high-precision meshes only in regions where detailed measurement is necessary, thus balancing measurement precision with computational complexity.

Inventive Principle:
Principle #1Segmentation

3Productivity

If automated mesh generation is implemented, then productivity increases, but control over mesh quality may be reduced

Engineering Contradiction:
Improvemesh generation efficiencyVSAvoidmesh quality control
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The system incorporates automated feedback mechanisms that continuously monitor mesh quality metrics during generation. The algorithm assesses mesh quality in real-time and adjusts generation parameters accordingly, ensuring high mesh quality is maintained while benefiting from automated high-speed generation. The feedback loop enables the system to self-correct quality issues without manual intervention.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP3837673B1Feature based abstraction and meshing
Publication Date: 2026.05.20 SIEMENS INDUSTRY SOFTWARE INC
  • EP3837673B1 patent drawingFigure 1A~1B
  • EP3837673B1 patent drawingFigure 2A~2B
  • EP3837673B1 patent drawingFigure 3A~3B

AI summary

Methods for CAD operations and corresponding systems (2800) and computer-readable mediums (2826) are disclosed herein. A method includes receiving (502) a model (600) of a part to be manufactured, wherein the model includes a plurality of original faces (102, 104, 106, 112, 114). The method includes classifying (510) each face in model according to a relative face curvature according to classifications that include at least a high-curvature classification (702). The method includes classifying (514) any sliver faces (102, 104, 106, 112, 114) and narrow blend faces (402, 404, 406, 408) of the plurality of faces. The method includes merging (516) contiguous faces (702) in each classification. The method includes detecting (518) special faces (1002, 1012) of the plurality of faces. The method includes restoring (520) original faces in the highcurvature classification except for the special faces (1002, 1012). The method includes processing (522) shared edges of the restored original faces to produce merged faces (802). The method includes merging together (524) any merged faces that produce a locally narrow face (302) or an isthmus (202). The method includes storing (526) a modified model of the part to be manufactured.