Manufacturing

Scan, repair, reprint: replacing obsolete and broken parts

November 11, 2026

The scan, repair and reprint route for obsolete parts, and the part nobody mentions: a printed replacement behaves differently from a moulded one.

Illustration of a scanned gear with two broken teeth rebuilt and highlighted in yellow, ready to reprint

TL;DR: Scan the part, repair the geometry, reprint it. The route is well proven and the lead time savings are real, but the part that comes back behaves differently from the moulded original. A printed replacement typically reaches 70 to 90 per cent of moulded strength along the layers and much less across them, so orientation and material choice decide whether it lasts. This post covers all three stages and the parts that don't suit it.

A bracket snaps inside a packing line. The machine is nineteen years old, the manufacturer stopped supporting it in 2014, and the part number returns nothing. Somebody suggests 3D scanning and printing services, and the obvious question follows: does that give you a real part or a plastic imitation of one?

Both, depending on how it's done. The route works, and there's hard evidence behind it. Shell Nigeria reverse engineered and printed an obsolete seal cover for an offshore mooring buoy, cutting the lead time from sixteen weeks to two and the replacement cost by 90 per cent.

What decides the outcome is the middle stage. The scan is simple and the print is quick. Everything worth knowing happens in between, and the honest version includes what a printed part gives up.

What does scan, repair, reprint mean?

Three stages. Capture the broken or worn part as measured geometry, rebuild that geometry into something sound and printable, then manufacture the replacement. The output is a physical part and a file, and the file is often worth more than the part.

It's used where the original supply has gone. Industrial machinery, transport and energy assets run for decades, and by then tooling has been scrapped, suppliers have disappeared, and requalifying production for a handful of units costs more than anyone will pay. The driver is uptime rather than unit cost.

The file is the second prize and the larger one. Obsolete parts can sit in a virtual catalogue and be printed at the point of need instead of occupying a warehouse. Our 3D scanning service is the front end of that, and the model comes back to you at the end of it.

Capturing a part that's already broken

Every part that arrives here has failed at something, and that changes the capture.

Worn surfaces are the first problem. A gear that's been running for eleven years is no longer the gear that was designed, and scanning it faithfully gives you the worn version. Somebody has to decide which surfaces represent the design and which represent eleven years of service. The unworn side of a symmetrical part answers that quickly. A mating component answers it better.

Missing material is the second. A snapped bracket with half the mounting face gone leaves a gap nothing can measure. Automatic hole filling will close it, but the new triangles simply follow the curvature of the surrounding mesh and may not match what was there. That's a guess with a smooth finish on it.

Which is why we ask for everything. Both halves, the worn one and the spare one, the part it bolts to, and a photo of the assembly. More reference means fewer assumptions, and where an assumption is unavoidable we write it down rather than print it.

What does repair mean at this stage?

Two different jobs share the word. Mesh repair makes the scan printable by closing holes and fixing bad geometry. Design repair fixes the fault that broke the part in the first place. The first is housekeeping. The second is engineering, and it's the one worth paying for.

Mesh repair deals with a known list of faults: non-manifold edges, holes and flipped normals, capped and stitched until the mesh is watertight. Slicing software rejects anything that isn't. Plenty of it is automated now.

What isn't automated is changing the shape. As one guide puts it, repairing a mesh so it prints and editing a mesh to change its shape are different jobs, and the second is much harder without the original CAD. That's the argument for taking the scan into reverse engineering rather than printing straight from it.

Design repair is where our product design team earns its keep. If a part snapped in the same place every time, that section gets thicker. If a rib was too thin for the load, it gets redrawn. Replacing a part exactly as it was means replacing its weakness too.

How does a printed part differ from the moulded original?

It's directional. A moulded part has similar strength whichever way you pull it. A printed part is built in layers, and the bond between layers is weaker than the material within them. Expect 70 to 90 per cent of moulded strength along the layers and considerably less across them.

The published figures line up. Testing puts FDM parts at 70 to 90 per cent of injection-moulded strength in the XY plane, with the Z axis at 40 to 75 per cent as strong and only 10 to 30 per cent as ductile, and long-standing work on ABS found printed specimens reaching 65 to 72 per cent of moulded tensile strength depending on how the material was laid down.

The reason is physical. Layers bond by partially remelting the layer below, and that weld is never quite as strong as the plastic either side of it. Resin printing behaves differently: testing on stereolithography parts found maximum stress holding constant across build angles, which makes it the better choice where a part gets loaded from an awkward direction.

We run FDM and mSLA, so the process gets picked against the part rather than against what's free that afternoon. That's part of what industrial 3D printing means as opposed to putting a file on a machine.

Which way up? Orientation and load

Orientation is a design decision that happens after the design is finished, and getting it wrong wastes everything upstream.

The rule is simple to state. Layer lines should run along the load, not across it. FDM parts are often four to five times weaker in tension along the Z axis than in the XY plane, so a part loaded the wrong way relative to its layers fails at a fraction of what it should carry.

One reported example makes it concrete: a hook specified to hold 10 kg was printed flat on the bed so the load pulled the layers apart, and it snapped at 3 kg. Same file, same material, same machine. Different orientation.

Smaller settings pull in the same direction. Thicker layers mean fewer bonded interfaces and better adhesion across them, hotter extrusion improves the weld, and infill density carries load that thin walls can't. None of it rescues a part oriented against its own load path.

Which parts suit this, and which don't

Good candidates are discontinued, needed in small numbers, and holding moderate loads at ordinary temperatures. Poor candidates are safety-certified, highly stressed, running hot, or sitting in something aggressive. The economics work when tooling would cost more than the parts are worth.

Industry guidance draws the line in similar terms: printing suits parts that are discontinued, needed in fewer than about fifty units a year, urgent on lead time, or where tooling costs exceed the total value of the parts. Housings, brackets, covers, clips, spacers, guides, handles and trim all sit comfortably inside that.

Outside it: anything with a safety certificate behind it, pressure-containing parts, and components under constant high load. A geometric copy is not a qualified replacement, and we'll say so at the quote stage rather than halfway through.

The economics can still surprise you at the small end. A UK equipment supplier printed a replacement bracket for its label printers that couldn't be sourced anywhere, avoiding the cost of replacing the printers themselves. One plastic part, several thousand pounds of avoided spend.

Two things to take away. The scan and the print are the quick parts; the decisions in between about wear, missing geometry and orientation are what make the replacement last. And the file you get at the end turns the next failure into a print job rather than another investigation.

Send us the part, both halves if it's in two, along with anything it bolts to. There's no minimum order quantity, the minimum order value is £50, and most orders leave in 3 to 5 working days once the design is signed off. Email sales@nexform.co.uk or use the contact form, and we'll quote the job through to low volume manufacturing if you need more than one.

Written by

Scott Watts

Company Owner (Operations Manager)

Founder and owner of Nexform, leading every project from first enquiry to finished piece.

faq.

Can you recreate a part that's snapped in half?

Often, yes, though it depends on what's missing. If the break is clean and the geometry can be worked out from the other side, symmetry or a mating part, we can rebuild it. If a whole section has gone and nothing tells us what shape it was, we'll say what we're assuming rather than filling the gap quietly. Send every fragment you have.

Will a printed replacement be as strong as the original?

Not in the same material, and not always in the same way. A printed part typically reaches 70 to 90 per cent of moulded strength along the layers and considerably less across them. Orientation, material and wall thickness close most of that gap on most parts. Highly stressed and safety-critical components are a different conversation.

How quickly can I get a replacement part?

Most orders leave our workshop in 3 to 5 working days once the design is signed off. Scanning and modelling happen first, and complex parts take longer at that stage. You get a date with the quote rather than an estimate after the fact.

Can you make just one?

Yes. There's no minimum order quantity and the minimum order value is £50, so a single replacement is a normal job for us. Once the CAD model exists, reordering later costs a fraction of the first one because the design work is already done.

What happens if the part breaks again?

The part goes on file, so the second batch comes off the same model with no repeat design cost. That's the real value of the exercise: you've swapped a part you can't buy for a file you own, and the next failure is a print job rather than a research project.

What should I send with the part?

Send the mating parts, or photographs and dimensions of what it fits into. A replacement that matches the old part perfectly still fails if the old part had worn or distorted. Checking against the assembly is how we catch that before anything gets printed.

related posts

get in contact.

have an enquiry?
send a message.
Any file type, up to 500 MB
Message sent.


We will be in touch shortly.