A web-based structural engineering tool that lets civil engineers draw arbitrary steel cross-sections and instantly compute their structural properties — built during a 4-month internship and currently used by engineers at the company.
Status: Deployed and in active use by employees within the company
Civil engineers working with complex steel cross-sections need to know properties like moment of inertia, section modulus, and radius of gyration before they can design safe structures. Calculating these manually for non-standard or compound sections is tedious and error-prone.
This tool lets an engineer draw any steel cross-section directly in the browser, and instantly get all structural properties computed in real time.
- Draw I, C, and L steel cross-sections in 4 orientations each
- Combine multiple sections into complex compound cross-sections
- Zoom, pan, snap, undo, and right-click dimension editing
- Real-time computation of:
- Centroid
- Cross-sectional area
- Second moment of area (Ixx / Iyy)
- Section modulus
- Radius of gyration
- Standard steel profile database (ISJB, ISLB, ISMB series) for quick insertion of standard shapes
Built entirely from scratch using the HTML Canvas API — no drawing libraries. Custom geometry functions handle each section type with proper arc fillets, flange/web ratios, and direction-aware orientation. Mouse event handling covers dragging, snapping (default and custom modes via Shift key), double-click selection, and a full undo stack.
A FastAPI server uses Shapely for geometry and triangle/pygmsh for mesh generation. Shapes drawn on the canvas are serialized and sent to the backend, where they are triangulated and all structural properties are computed via numerical integration over the mesh.
Steel section dimensions for ISJB, ISLB, and ISMB series are stored in a SQLite database. To populate this database without manual entry, a separate OCR pipeline was built using Google Cloud Vision API — converting IS code specification PDFs directly into structured database entries via bounding box analysis and spatial row/column detection.
| Layer | Technology |
|---|---|
| Frontend | HTML, CSS, JavaScript, Canvas API |
| Backend | Python, FastAPI, Shapely, pygmsh |
| Database | SQLite |
| OCR Pipeline | Google Cloud Vision API, pdf2image, openpyxl |
This project was built during an internship. The source code is proprietary and not publicly available. The demo video above shows the fully working tool.