case study

Pressure Vessel Nozzle QA

As-built point cloud of a nozzle flange with the fitted nozzle axis used for deviation measurement.

On an offshore water-injection vessel fabrication project, I used terrestrial laser scanning to capture the as-built geometry of the vessel and its nozzles, then compared that point cloud against the as-designed CAD model to verify manufacturing quality before the vessel left the shop.

For each nozzle I extracted the flange-face center, fitted the nozzle axis, and measured elevation, radial position, and angular orientation directly from the point cloud. Each measurement was compared to the engineering drawings and flagged against tolerance bands — pass, watch, or fail — so the fabrication team could see at a glance where the build matched intent and where it drifted.

The analysis also accounted for data quality: highly reflective machined flange faces introduce noise into LiDAR returns, which can reduce confidence in the fitted geometry. Flagging those lower-fidelity regions kept the report honest about which deviations were solid measurements versus indicative ones.

The full pipeline — point-cloud registration, geometry extraction, tolerance analysis, and a branded PDF report — was built in R and Quarto, producing a reproducible deliverable that regenerates automatically when the underlying scan data is updated.