Why 3D Printed Parts Warp, Crack and Delaminate, and How to Stop It

A part lifts off the plate at the corners. A tall print splits horizontally halfway up. A large flat panel comes off the bed looking like a shallow bowl. A part that seemed fine develops a crack three days later, sitting on a shelf.

These look like four different problems. They're one problem, and understanding it makes all four predictable: and mostly preventable.

The root cause: plastic shrinks when it cools

Thermoplastic is deposited molten, at 220-300 °C depending on the material, and it cools to ambient. As it cools it contracts. Every material does this, and each has a characteristic shrinkage: PLA contracts about 0.3%, PETG around 0.5%, ABS and ASA closer to 0.8%.

If the whole part cooled at once and uniformly, this would be harmless. The part would come out slightly smaller than modelled, and you could compensate in CAD.

But a print doesn't cool uniformly. It's built one layer at a time over hours. The bottom layer has been cooling since the print began. The top layer is still molten. Every layer in between is at a different temperature, and each one is contracting at a different rate while being physically bonded to its neighbours.

That mismatch is a stress. Every layer is pulling on the layer below it, and the layer below is resisting. The stress accumulates as the part grows. Eventually, something gives. And what gives determines which of the four symptoms you see.

The four symptoms

Corner lift

The bed holds the bottom of the part down. As upper layers cool and contract, they pull inward and upward. The strongest pull is at corners, where material contracts along two axes at once.

If the accumulated pull exceeds the bed adhesion, the corner peels off the plate. Now the part is no longer flat, subsequent layers are printed onto a warped foundation, and the geometry is wrong from that point up.

Bowing

Same physics, but the adhesion holds. Instead of the corner lifting off, the whole part curls, the contraction bends it into a shallow dish. You get a part that's dimensionally correct in X and Y but not flat.

Layer separation, or delamination

Instead of the part deforming, the stress exceeds the bond strength between two layers, and they let go. You get a clean horizontal split, usually partway up a tall part and usually at a point where the cross-section changes.

This is the most dangerous failure because it's often invisible. The part looks fine, and the crack is a hairline you can't see until it opens under load.

Delayed cracking

The most unsettling version. The print completes, the part looks perfect, and days later there's a crack in it.

The stress was locked in during printing but hadn't quite exceeded the bond strength. Then something tipped it over: a temperature change, a small mechanical load, or the polymer relaxing over time. The failure was designed in during the print; it just took a while to show.

Why ABS and ASA are the usual suspects

Two properties compound.

Higher shrinkage. ABS and ASA contract roughly 0.8% on cooling, against PLA's 0.3%. Nearly three times the contraction means nearly three times the stress for the same geometry.

Higher print temperature. They extrude around 250-260 °C into a room at 25 °C. That's a 230 °C drop. PLA printing at 200 °C has less far to fall, and it falls more gently.

Polycarbonate is worse still on both counts. PLA and PETG are comparatively forgiving, which is exactly why beginners are told to start with PLA.

Fix one: an enclosed, heated chamber

This is the real fix, and it's why both of our machines are enclosed.

An enclosure traps the heat radiating from the bed and the part itself, raising the ambient temperature inside to perhaps 45-55 °C. That does two things. It reduces the temperature difference each layer must fall through, so there's less contraction stress. And it keeps the layers below the current one warmer and more compliant, so they can relax slightly rather than resisting rigidly.

The effect is dramatic. A large ABS part that will reliably warp off an open-frame printer will print flat and stay flat in a sealed chamber.

This is the single most important question to ask a print bureau about ABS, ASA or PC: is the machine enclosed? If it isn't, you'll get parts that warp, and no amount of adjusting other settings fully compensates.

It's also why draughts matter. An enclosed printer next to an open window or under an air conditioner still has a problem: a cold draught across the chamber produces exactly the uneven cooling the enclosure exists to prevent.

Fix two: design the stress out

Process control only goes so far. Geometry is often the real culprit, and these changes cost nothing.

Fillet or chamfer the bottom edges. A sharp 90° corner where the part meets the plate is the peak stress point and where lift begins. A modest fillet spreads that stress over a curve.

Avoid large flat bottom faces. A 200 × 200 mm flat plate is close to the worst possible geometry for warping, maximum contraction across the largest span, all of it fighting the bed. If your part has a large flat face, consider whether it needs to be one piece, whether it can be ribbed rather than solid, or whether printing it on edge would be better.

Keep the cross-section consistent. Delamination happens preferentially where the cross-section changes abruptly, because the stress distribution changes there too. A gradual transition is far less likely to split than a sudden step.

Reduce mass where you can. Less material means less total contraction. A part with sensible wall thicknesses and moderate infill warps less than the same part printed solid, and it's cheaper and faster too.

Round the corners in plan view. Sharp external corners concentrate the inward pull. Even a 3 mm radius helps noticeably on a large flat part.

Fix three: choose a material that doesn't do this

The most reliable way to avoid warping is to print something that barely warps.

If your part doesn't actually need ABS's heat resistance, PETG gives you most of the toughness with a fraction of the shrinkage and no enclosure requirement. If it doesn't need to be outdoors, you don't need ASA.

This is worth being honest about, because ABS and ASA get specified more often than they're needed. Ask what the part actually has to survive. If the answer is "it sits on a desk", PETG will do the job, print more reliably, and cost less.

Where you do need the heat resistance or the UV stability, use the right material and print it in a proper enclosure. Don't compromise by printing ABS on an open machine and hoping.

Fix four: process details that matter

These are ours to get right, but they're worth knowing about.

Bed adhesion. A clean plate, correct first-layer squash, and the right surface for the material. A brim, a single-layer skirt attached to the part's base, adds adhesion area at the corners where lift starts, and it peels off afterwards.

Bed temperature. High enough to keep the base compliant. For ABS and ASA that means 100 °C or so.

Part cooling fans. Counterintuitively, ABS and ASA want the part fan off or very low. Cooling fans exist to help PLA solidify quickly for detail, but on high-shrinkage materials they create exactly the rapid, uneven cooling that causes warping. PLA wants full cooling; ABS wants almost none.

Drying the filament. Nylon and PC absorb moisture from the air, and moisture in the filament flashes to steam at the nozzle, leaving voids that weaken layer bonds. PA-CF is dried for eight hours before every run here and ships with desiccant for a reason.

Diagnosing your own failure

If you've a warped or split part, work through this:

Lifted at the corners, flat elsewhere → bed adhesion insufficient for the contraction stress. Enclosure, brim, bed temperature, fillet the bottom edges.

Whole part bowed but still stuck down → contraction stress exceeded part stiffness. Reduce the flat span, add ribs, or split the part.

Clean horizontal split partway up → interlayer bond lost. Chamber temperature, cooling fans too high, or the cross-section changes too abruptly at that height.

Crack appeared days later → locked-in stress finally released. Almost always an enclosure or geometry problem, not a settings problem.

Only happens on big parts → confirms it's contraction stress, which scales with span. Small parts of the same material printing fine is diagnostic, not reassuring.

What we do about it

Every job is reviewed before printing, and geometry likely to warp gets flagged. Sometimes that means suggesting a fillet or a rib. Sometimes it means recommending PETG instead of ABS because the part doesn't need the heat resistance it was specified for. Sometimes it means splitting a large flat panel into two pieces that each print flat.

Both cells are fully enclosed, with the chamber sealed and the exhaust filter engaged for ABS and ASA. Bed temperatures, cooling profiles and adhesion are tuned per material rather than left on a generic preset.

If you've a part that keeps warping, whether we printed it or not, send us the geometry and describe the failure. Warping is a solvable problem, and the solution is usually a combination of a small design change and a properly controlled chamber rather than any single heroic fix.