Rapid prototyping with 3D printing

The phrase rapid prototyping sets an expectation the process does not meet on the first attempt. A part is designed, printed, and does not fit. That is not a failure of the method; it is the method. The value is in how cheaply you can be wrong and how quickly you can be wrong again.

What follows is what the loop actually looks like, where the time really goes, and what changes when a prototype has to become a product.

The loop

Model, print, measure, revise. Repeat. For a part with any kind of fit, expect three to five passes before it is right. For a part that stands alone with no mating surfaces, one or two.

Almost all of those revisions are dimensional rather than structural. The shape was right; a hole was 0.3 mm too small, a lip fouled a corner, a wall was too thin where a screw went in. This matters, because it means the thing worth optimising is how fast you can change a number and get another part, not how fast you can design from nothing.

Print the question, not the part

The most common waste in prototyping is printing the whole part to answer a question about one face of it.

If this round is about whether the mounting holes line up, print a 3 mm plate with the hole pattern on it. Twenty minutes instead of six hours, and it answers exactly the same question. If it is about whether the clip engages, print the clip and the 20 mm of wall it clips into.

Full parts are for when the question is about the whole part: does it feel right in the hand, does it survive being dropped, does everything assemble together. Those are real questions and they deserve the six hours. Most rounds are not those.

Measure properly and write it down

Take digital calipers to every feature that matters and record two numbers: what you modelled and what you got. The difference is the interesting part.

That difference is largely a property of your printer and your material, not of the individual part, so it stays roughly constant. Once you know that your machine prints holes about 0.25 mm undersize and outside dimensions about 0.1 mm oversize, you can apply that knowledge deliberately instead of rediscovering it every time. This is the difference between three iterations and six.

Revise with the measurement, not the impression. A hole that came out 0.3 mm undersize needs 0.3 mm added to that hole. Scaling the whole part to 101 percent fixes the hole and breaks everything else. Design for 3D printing covers the baseline allowances to start from.

Where the time actually goes

People assume the printer is the bottleneck. It usually is not. A small part prints overnight while nobody is waiting for it, and the machine does not need supervising.

The delay is the CAD revision, and specifically the wait for someone to sit down and do it. If that person is you and you have other work, the model waits until Thursday. If it is a colleague, it joins a queue. The print took four hours; the round took four days.

This is the honest case for generating a starting model from a description: not that it produces better geometry than an engineer would, but that it removes the queue for the first version and for the routine dimensional revisions. For a bracket, plate, adapter or enclosure, that is often most of the loop. For anything where the shape depends on judgement rather than description, it is not, and text to CAD vs traditional CAD sets out where that line falls.

What a printed prototype does and does not tell you

It answers form and fit well. Does it assemble, does it clear the other components, is it a sensible size in the hand, does the cable reach. These are the questions most iterations are about, and a printed part answers them properly.

It answers function only partially. A printed part is anisotropic: markedly weaker across its layers than along them, so it fails in ways the final part will not, and survives loads the final part would handle differently. Treat a strength result from an FDM prototype as directional rather than a number you can design to.

It does not answer surface finish or long-term behaviour. Layer lines are a printing artefact. Creep, UV degradation and fatigue on a printed polymer tell you very little about a moulded one.

When to move off FDM

Change process when the question changes, not on a schedule.

• Resin, meaning SLA or MSLA, when you need fine detail or a smooth surface: small features, threads that must engage, anything being shown to a customer. It is more brittle and needs washing and curing, so it is worse for functional testing.
• SLS when you need genuinely functional nylon parts. Isotropic, tough, no support marks because the powder supports the part. It is the step where prototypes start behaving like production parts.
• CNC machining when the material properties themselves are the question. If you need to know how it behaves in aluminium, no printed polymer will tell you.

One rule worth keeping: change the design or change the process, not both at once. If you do both and the result is worse, you have learned nothing about which change caused it.

Prototype to production

This is where a printed design stops transferring, and the extent surprises people. A part designed for FDM and a part designed for injection moulding are different parts.

• Draft angles. Every face parallel to the mould direction needs a degree or two of taper so the part releases. Printing needs none of this, so printed designs never have it.
• Uniform wall thickness. Thick sections cool slowly and leave sink marks and voids. A solid block that prints fine must become a shell with ribs.
• Ribs instead of bulk. Stiffness comes from geometry rather than material, typically ribs around 60 percent of the wall thickness to avoid sinking.
• Gate and ejector positions. Both leave marks. Decide where they are acceptable before the tool is cut.
• Different tolerances. Moulded parts shrink predictably by material. Printed allowances do not transfer.

None of this makes the printed prototype wasted. It answered form and fit for a few pounds of filament, which is what it was for. Just do not expect to hand the same file to a moulder.

Costing an iteration honestly

The material cost of a prototype round is close to noise. A palm-sized part is a small fraction of a spool.

The real cost is elapsed time, and it is dominated by the human step. If a round takes three days and two of them are waiting for CAD attention, then halving print time changes almost nothing and removing the CAD queue changes almost everything. Measure your own loop before optimising the part of it that is easiest to see.

Where to go next

Design for 3D printing has the clearances and wall thicknesses that remove one or two iterations before you start. STL files explained covers why you should keep the STEP alongside every printed revision. For the tools themselves, see CAD software for 3D printing, and the mechanical parts in the gallery show what a described part looks like alongside the description.