What reverse engineering a spare part really involves
Reverse engineering has a reputation problem on the maintenance side of the business, and it is partly deserved. Plenty of it is done badly by a workshop that measures a worn sample, machines to nominal, and ships it. The result fits, runs for a few months, and fails in a way that is hard to trace back.
Done properly, reverse engineering is a legitimate engineering activity with a clear scope, a defined measurement method and an acceptance test. This article sets out when it is worth doing, what information you need before you start, and where the real risks sit.
When reverse engineering is the right answer
It becomes attractive in a specific set of circumstances, and unattractive in others.
| Situation | Reverse engineering | Better alternative |
|---|---|---|
| Simple mechanical part, no safety function | Good fit | — |
| Original maker still exists, part is expensive but available | Poor fit | Buy original, negotiate price |
| New-old-stock exists in the secondary market | Usually unnecessary | Source and verify NOS |
| Part carries a pressure, lifting or safety certification | Avoid unless you can recertify | Original or OEM-approved equivalent |
| Electronics with firmware or calibration data | Very high risk | Repair, or migrate to a modern equivalent |
| Machine is the only one of its kind and stopping is not an option | Acceptable with testing | — |
The pattern is straightforward: reverse engineering works best on geometric parts with well-understood materials and no regulatory obligation attached to them. It works worst where a certificate, a firmware image or a safety case is involved.
The information package you need before quoting
A workshop cannot reverse engineer a worn part in isolation. The minimum package is:
- The failed part itself, plus a second sample if the part is still running elsewhere on site. Two samples let the workshop separate the original geometry from wear.
- A nameplate photograph from the machine, showing the maker, model and serial. This is what allows anyone to trace the original specification.
- Any original documentation — a drawing, a parts list page, an installation manual, even a photograph of the page in a printed manual. Drawings sometimes surface through the machine’s documentation even when the part is long gone.
- Rest of the assembly, if the part mates with anything precision. Measuring a shaft without measuring the bore it runs in tells you half the story.
- Operating conditions — temperature, load, cycle rate, media, lubrication. Two parts can look identical and be specified completely differently because one runs in a 90 °C gearbox.
Photographs are cheap and remove a surprising number of false starts. Take them against a plain background with a ruler in frame, from at least four angles, with close-ups of any markings, wear surfaces and damage.
What “measuring” the part actually means
The single most common technical error is measuring a worn part and treating the result as nominal. Wear removes material; measurement of a used part gives you a value that is smaller than nominal in exactly the places that matter most.
Workable practice is to measure critically worn features on a brand-new reference if one can be found, and otherwise to infer nominal from the interface: measure the mating bore, the shaft the part rides on, the bore spacing on the housing. Interfaces usually outlive the consumable part, which makes them a better source of truth.
Two other things worth stating up front in the drawing package:
- Tolerances must be specified, not implied. A dimension with no tolerance will be made to whatever the workshop finds convenient.
- Material and heat treatment are part of the specification. A gear cut from the right geometry and the wrong steel fails in weeks. If the material is unknown, ask the workshop to test a sample rather than assume a common grade.
Acceptance testing: agree it before you order
Decide how the part will be accepted before the first item is made, and write it into the purchase order. Options that work in practice:
- Dimensional report — first article inspection against the drawing, with actual measured values rather than a tick box.
- Material certificate — mill certificate for the steel, hardness test results after heat treatment.
- Bench or no-load run — for rotating parts, a run-in period with vibration and temperature recorded.
- Service trial — one unit installed on a monitored asset, with the rest of the batch held until it completes an agreed number of hours.
The last option is the one that actually derisks the project, and it is the one most often skipped because it delays the quantity saving. Holding back 80% of a batch for two weeks is cheap insurance against 80 units of scrap.
Documentation you should keep
Once the part is made, the drawing and inspection records become the new baseline. Keep them with the machine documentation, note the deviation from original, and record which units are running which versions. Six years later, when a second batch is needed, that record is the difference between a two-week order and starting over.
How we help
We work with Chinese manufacturers daily and can take a reverse-engineering request from sample photographs through to a first article, including material verification and inspection reports. Send the nameplate photograph, the failed sample and the operating conditions, and we will tell you honestly whether reverse engineering is the economical route or whether you should be sourcing the original.
Need a specific reference? Contact us freely at zhuaotech.com/contact-us — send the part number and we will confirm what we can supply.
Related reading
- Sourcing obsolete and discontinued industrial parts: a practical playbook
- Last-time buy planning: how to estimate lifetime demand
- How to describe a spare part you cannot identify
Need a specific reference? Contact us freely at zhuaotech.com/contact-us — send the part number and we will confirm what we can supply.