Automating P&ID Annotation Across 100+ Drawings

Challenge

The client, a mid-sized EPC firm, was tasked with reviewing and annotating over 100 vendor-supplied P&IDs as part of a fast-track project. These drawings contained a wide range of instrument tags, inline elements, and piping arrangements — but arrived in scanned PDF format with no editable metadata.

Manual annotation across this volume would have taken several engineers multiple weeks — with high chances of inconsistency, error, and duplicated effort.

Solution

We deployed the Storm Consulting eAI Desktop App on the client’s local systems. The tool was configured to:

• Run fully offline (air-gapped network)
• Detect standard instrument tags, line connections, and inline elements using OpenCV + OCR
• Apply optional OpenAI support (via anonymized prompt logic) to correct malformed tag IDs
• Output annotated data in both image and structured CSV formats for QA and MTO workflows

Implementation Highlights

• Zoom-level tuned OCR for precise tag detection across varied resolutions
• Tag normalization logic to standardize readings like A1-PG-25 and AI-PG-25
• Selective human review checkpoints (only 15–20% of tags required manual validation)
• Final dataset included ~6,200 tags extracted and classified from 100+ P&IDs
• Project completed in 4 working days with just 2 engineers for review

Results

“The speed and traceability this tool offered were unmatched. It didn’t just reduce work — it gave us confidence in the data.” — Lead Process Engineer, Client Team

Why It Worked

• Local-first architecture met the client’s IT/security requirements
• Output structure mimicked existing workflows (image + spreadsheet)
• Optional AI inference offered smart corrections without enforcing black-box decisions
• Minimal training needed — engineers were productive within 15 minutes

Share

Like this? Share it with your network!

• Share on Facebook
• Share on Twitter
• Share on LinkedIn