PSLshell

Topology-aware principal stress line tracing on shell structures.

PSLshell is a MATLAB implementation for tracing and visualizing principal stress lines (PSLs) on shell structures.

It accompanies the paper:

Junpeng Wang, Yingjian Liu, Jun Wu, and Rüdiger Westermann
Topology-aware Stress Analysis in Shell Structures
Computer Methods in Applied Mechanics and Engineering, 2026

---

Overview

PSLshell performs topology-aware stress analysis on shell meshes and extracts:

• Major and minor principal stress lines (PSLs)
• Degenerate points of the stress field
• Topological skeleton structures

The method operates directly on FEM stress fields and supports mixed shell meshes.

---

Features

• Principal stress line tracing on shell surfaces
• Supports shell elements:
  • T3 / Q4 / T6 / Q8
• Works with triangular, quadrilateral, and hybrid meshes
• Handles node-wise or element-wise stress data
• Topology-aware PSL seeding and distribution control
• Detection of degenerate points and skeleton extraction
• Lightweight .TSV interface for FEM data
• Single-file MATLAB implementation (PSLshell.m)

---

Input Data

PSLshell reads stress fields from a .TSV file, which contains:

• Mesh vertices
• Element connectivity
• Boundary condition Info (optional)
• Stress tensor components

The stress tensor can be defined in:

• Local shell coordinates, or
• Global Cartesian coordinates

---

Quick Start

Run in MATLAB:

PSLshell

Key Parameters

• PSLsDensityCtrl
  Controls PSL density (larger → denser lines)

• PSLtypeCtrl = [major, minor]
  0 = generate, 1 = skip

• topologyAware
  Enable topology-aware PSL tracing

• distanceMetric
  'Euclidean' or 'Geodesic'

• seedSparsityCtrl
  Controls seed sampling density

• maxIts
  Maximum integration steps per PSL

Output

PSLshell generates:

• Major PSLs
• Minor PSLs
• Degenerate points
• Topological skeletons

Citation

If you use this code, please cite:
@article{wang2026topology,
  title={Topology-aware stress analysis in shell structures},
  author={Wang, Junpeng and Liu, Yingjian and Wu, Jun and Westermann, R{\"u}diger},
  journal={Computer Methods in Applied Mechanics and Engineering},
  volume={452},
  pages={118770},
  year={2026},
  publisher={Elsevier}
}

Author

Junpeng Wang
Technical University of Munich
junpeng.wang@tum.de