When the original supplier is gone, the broken part becomes the specification. The trick is knowing what to copy — and what to correct.
The scenario is always the same
A fifteen or twenty year old machine breaks down. The part is out of catalogue, the manufacturer has been bought out, the drawings were never delivered. The quotation for full machine replacement arrives at a figure unrelated to the actual problem, and a lead time unrelated to the production season.
Copying is the wrong question
The part in front of you is worn. Digitising it as-is reproduces the wear: the clearance that caused the failure is carried into the new part. So the work starts by separating functional geometry from degraded geometry, which means understanding what the part does, not only what it looks like.
From point cloud to a model you can decide on
A scan mesh is not a CAD model. It cannot be modified, carries no design intent and does not let you adjust a dimension. Parametric reconstruction recovers the features — axes, datum planes, fillets, threads — and makes the part editable, therefore improvable.
Validate before running the series
A refabricated part is validated on three fronts: dimensionally, against the model; materially, by certification or analysis; functionally, by fitting and testing. Skipping the third is the most common way to lose two months.
The real gain is not the part
Once the parametric model exists, it stays. The next failure does not restart the process: it triggers an order. That is where reverse engineering stops being an emergency operation and becomes a spare parts policy.
When it is not the right answer
If the part is available off the shelf, cheap and quick to get, there is no reason to remake it. Reverse engineering earns its place on unavailable parts, long lead times, or obsolescence that threatens continuity of operations. We say so before quoting.
Published 11 June 2026 Discuss this topic with us
