Power Curves Trapped in Drawings

The situation

The power-curve data for turbines and inverters - the numbers that describe how a machine performs - existed only inside issued-for-construction drawings. It was there, technically, but locked in a format no analysis could reach. Anyone who needed a curve had to find the right drawing and read it off by eye. This was the first project the current delivery team took from start to finish on its own.

The defect

Knowledge stuck in a PDF is knowledge you cannot use. A power curve buried in a construction drawing cannot feed a performance model, a report, or a comparison across machines without someone first transcribing it by hand. The data existed but was effectively offline. And the variability between drawings meant no two were laid out quite the same, which is exactly what makes reading them by hand both slow and easy to get wrong.

What we engineered

• Built extraction that reads issued-for-construction drawings and produces usable power-curve data from them.
• Handled the variability between drawings, since real documents rarely follow one clean template.
• Hardened the tool in a second generation after user feedback exposed the messier inputs.
• Output curve data in a form that downstream models and reports can actually consume.
• Freed engineering knowledge that had been locked in PDFs since the site was first built.

The result

Power curves that used to live as pictures in a drawing are now data an engineer can actually work with. The second generation, shaped by real user feedback, handles the variability that makes these drawings awkward to read by hand. As the first project the current delivery team closed end to end, it also set the pattern SCADADOG keeps applying: engineering knowledge locked in documents stays unusable until you get it back out.