Industrial data diagnostic
Plant 3D P&ID Data Reconciliation Diagnostic
A read-only diagnostic that reconciles archived logical-line records with live Plant 3D P&ID-linked pipeline objects and exposes drill-down evidence, drawing traceability, and bounded review queues.
What this project proves
- Six review workspaces and seven bounded reconciliation queues
- Five-step path from command-centre signals to source-linked evidence
- Deterministic classifications for archive-only, live-match, drawing, specification, fluid, endpoint, and placeholder signals
- Zero browser-to-database connectivity and zero source write paths
The problem
Plant engineering records can be split across archived line registers, drawing-linked Plant 3D data, and separate review workflows.
High-level reconciliation counts are not actionable unless engineers can inspect the exact logical-line population, linked drawings, and source objects behind each indicator.
My approach
Build six coordinated workspaces covering command-centre analysis, reconciliation, data quality, source intelligence, reporting, and drawing traceability.
Guide users through a five-step review path from KPI-level signals to exact logical-line evidence and drawing-linked live objects.
Protect the source through fixed parameterized SQL Server queries, typed route contracts, bounded request parsing, and explicit static-demo controls.
Architecture and engineering choices
- Next.js application routes with typed request and response contracts
- Fixed query registry and bounded URL/input parsing
- Read-only parameterized SQL Server access
- Explicit static demonstration mode that never silently replaces live data
Truthful scope and boundaries
- The application does not write to Plant 3D or SQL Server.
- It does not approve drawings, manage formal revisions, make engineering dispositions, or determine turnover readiness.
- The live public demonstration uses captured synthetic or sanitized evidence.