How strong is a 3D printed part, really?

People arrive at this question from two directions. Some assume printed plastic is a prototype material that will snap the first time it is loaded. Others assume that because a part exists, it is as strong as the injection-moulded one it replaced.

Both are wrong, and the reason is the same: a printed part does not have one strength. It has a different strength in each direction.

Why does direction matter so much?

A printed object is built in layers, each one welded to the layer beneath while the plastic is still hot. Within a layer the material is continuous and behaves close to the numbers on the filament datasheet. Between layers it is a bond, and a bond is always weaker than solid material.

The practical effect is large. Pull a part along the direction the layers run and it performs as expected. Pull it *across* the layers, trying to peel them apart, and it can fail at a fraction of that load.

This is why the same part, same material, same settings, can be either fine or useless depending on which way up it sat on the bed. It is also the single most common reason a part printed elsewhere arrives at our workshop broken.

So what makes a part strong?

Four things, roughly in order of how much they matter.

Orientation. Print so the layers run across the load, not along the direction it wants to split. This costs nothing. It is purely a decision made before the print starts, and it is the difference between a part that survives and one that does not.

Wall count. The solid perimeter shells carry most of the load in a printed part. Adding walls does more for strength than filling the inside, and costs less material.

Infill. The internal lattice matters less than people expect. Going from 20% to 50% adds weight, time and cost for a modest gain. Going from two walls to four usually achieves more.

Material. Last, not first. PETG and ABS are tougher than PLA in impact; TPU flexes instead of breaking. But a well-oriented PLA part beats a badly oriented ABS one, comfortably.

Which material would you actually choose?

For parts that do a job rather than sit on a shelf:

• PETG — the default for functional parts. Good impact resistance, tolerates heat and moisture, forgiving
• ABS — better under sustained heat and mechanical stress, for enclosures and mechanical components
• ASA — the same toughness plus UV stability, for anything living outdoors
• TPU — where the part needs to bend, grip or absorb a shock rather than resist it
• PLA — rigid and dimensionally precise, best for display, prototypes and indoor pieces

The full side-by-side, with tensile figures and heat tolerances, is on our materials page.

Can a printed part replace a broken original?

Usually, and this is a large part of what we do. Discontinued fittings, obsolete brackets, a clip that snapped and is not sold separately — these are good candidates, because the original was often over-specified for a job it never really had to do.

Two things make the difference between a replacement that lasts and one that fails again.

The first is understanding why the original broke. If it snapped from fatigue after twenty years, a like-for-like copy is fine. If it broke in six months because the design was thin at a stress point, copying it faithfully reproduces the flaw. We would rather thicken that section and tell you we have changed it.

The second is knowing the load and the environment. A bracket holding a shelf and a bracket holding an engine component are the same shape problem and completely different engineering problems.

When is a printed part the wrong answer?

We would rather say so early than quote and disappoint. Printing is a poor choice when:

• The part is safety-critical and failure has real consequences
• It carries a sustained heavy structural load
• It runs hot — near an engine, in a flue, against anything above roughly 100°C
• It needs certification we cannot provide, or must meet a documented material specification
• The volume is high enough that moulding is genuinely cheaper per unit

That last one is worth saying plainly, because it goes against our own interest: past a certain quantity, injection moulding wins on cost. We will tell you where that line falls for your part rather than let you find out afterwards.

How do you know it will hold before you commit?

Print one and break it.

For anything where strength matters, we would rather make a single part first and have you test it in the real position, under the real load, than deliver fifty and discover the design was marginal. A prototype costs a fraction of a production run and answers the question definitively.

If it fails, the way it fails tells us what to change — usually orientation or wall count, occasionally the material, sometimes the geometry itself.