Reverse engineering a part means measuring a physical object and rebuilding it as a clean, editable, manufacturable CAD model — and the hard part is not the scanning, it is the modeling judgment that turns measured points back into design intent. A 3D scan gives you a mesh of the part as it exists today, worn and warped included. A manufacturable model gives you nominal dimensions, standard hole sizes, tolerances, and features a machine shop can quote. This article walks through both halves of the job.

Three reasons people need it

  • An obsolete spare part. Something broke on aging equipment and the original maker is gone, or no longer stocks the part.
  • Lost source files. A product was designed years ago, the native CAD is gone, and you now want to improve it or move it to a different manufacturer.
  • Teardown analysis. Understanding how an existing product is constructed, usually to inform your own design.

All three demand the same engineering skill. Their legal positions are completely different, which we cover below.

How the part gets measured

Hand measurement

Calipers, micrometers, pin gauges, radius gauges, thread gauges. For prismatic and cylindrical parts — shafts, bushings, brackets, threaded fittings — hand measurement is still the most accurate and by far the cheapest route, because it produces a clean parametric model directly instead of an approximated surface. A turned part with bores and threads is almost always better reconstructed by hand than by any scanner.

3D scanning

Essential for freeform geometry: ergonomic grips, castings, organic shapes, sheet metal with complex draws. Structured-light and laser scanners output a point cloud or mesh that you then surface. The limitation to internalize is that a scanner measures the part as it is — including wear, deformation, shrinkage, and the dent from the day it failed — not as it was designed.

Supporting methods

Industrial CT or X-ray reveals internal passages and wall thickness you cannot reach with a probe. A coordinate measuring machine gives traceable high-accuracy data on critical features. And material identification — alloy or polymer analysis, hardness testing, sometimes plating thickness — is not optional: a part reproduced with perfect geometry in the wrong material fails in exactly the same place as the original. Our guide to choosing between aluminum and steel covers how to reason about a substitution when the original alloy is unavailable.

From measurement to model: where the real work is

The common misconception is that the scan is the deliverable. In practice a raw mesh is close to useless for manufacturing: you cannot parametrically edit it, you cannot cleanly change a fit, and no machinist wants to program from it.

The actual work is reconstructing design intent. The engineer recognizes that a slightly curved surface was meant to be flat, that a bore measured at 9.87 mm was drawn as 10 mm with a tolerance, that three holes sitting at odd angles were meant to be on a bolt circle, and that a fillet was a standard tool radius. The output should be a feature-based solid model plus a drawing carrying tolerances and finishes — which is why GD&T fundamentals matter as much here as in new design. Design-for-manufacturing thinking has to be applied to the reconstruction too, not simply inherited from a part that may have been made by a process you are not using.

Reproduce or improve?

A reverse-engineered part can almost always be better than the original. If it broke, there was a reason: a stress concentration at a sharp internal corner, a section too thin for the load, a material that embrittled with heat or UV exposure. Faithful reproduction faithfully reproduces the failure. We normally recommend a short stress review of the rebuilt part and consideration of local reinforcement — a larger fillet, a rib, a different alloy or temper — while verifying that nothing changes the interfaces to surrounding hardware. Fit, mounting, and clearance must stay identical unless you control the whole assembly.

What it costs, and when not to bother

A simple mechanical part is usually a few hours of measurement and modeling — low hundreds to low thousands of dollars in US engineering time. A complete assembly with dozens of parts is a multi-week project in the five-figure range. The biggest cost driver is not part count but how much of the geometry is freeform, plus the accuracy the application actually demands. Add fabrication on top: see CNC machining for prototypes for the making side, and CAD design services costs for the modeling side.

Ask the economic question first. A close-enough catalog part is frequently an order of magnitude cheaper than exact reconstruction, and for a single spare it often wins outright.

The legal side, briefly and clearly

Projects House is an engineering firm, not a law firm, and nothing here is legal advice — ask a qualified IP attorney about your specific situation. That said, the practical landscape: reproducing a spare part for your own use in equipment you own is generally uncontroversial. Copying someone else's commercial product and selling it is not. Between those poles sits a wide range that depends on live utility patents, design patents covering appearance, trademarks on shape or branding, and copyright on firmware and documentation. The workable rule is to check for active protection before reproducing anything commercial — start with a prior art search and, if you intend to sell, a freedom to operate search. Remember that appearance and function are protected separately, so a shape can be covered even when the mechanism is not. Learning from a competitor's product to develop something of your own is entirely legitimate.

Get your part measured and modeled properly

Projects House rebuilds physical parts into production-ready CAD and drawings for US clients — measurement, parametric modeling, material identification, and manufacturing documentation, including improvements where the original design was the problem. Tell us what the part is and what you need to do with it through the contact form, and we will tell you which measurement approach fits and what it will take. Our full mechanical engineering guide covers the surrounding topics.