A horizontal split appears part way up a tall part, usually on a corner or a flat face, often wide enough to see daylight through. The print carries on above it, so the finished object looks intact from some angles and is in two pieces from others.
The timing separates this from its neighbours. This happens during the print. A part that breaks later under load has weak layer bonding; a part that cracks overnight on the shelf is post-print stress relaxation.
Two forces pulling in opposite directions
Every layer contracts as it cools. The layers beneath it finished contracting minutes ago and are dimensionally fixed. So each new layer is laid down and immediately tries to shrink onto a base that will not shrink with it, which puts the whole stack into tension — the same mechanism that produces warping at the plate.
Meanwhile, the strength holding the stack together is the weld between adjacent layers, and that weld only forms while the previous layer is still hot enough for polymer chains to cross the boundary.
A crack appears where tension wins. On a material that contracts hard and welds only when warm — which describes ABS, ASA and polycarbonate exactly — the margin between those two quantities is small, and anything that cools the part faster erases it.
Why it happens at a height rather than everywhere
This is the detail that makes the fault diagnosable.
Near the plate, the bed radiates heat upward and the part sits in the warmest air in the machine. As the print grows, the top of it moves further from that heat source, into air that is several degrees cooler, and the previous layer is therefore colder when the new one lands on it. Somewhere up that gradient the weld stops being strong enough for the accumulated tension, and that is the height at which the part splits.
It is why the same file prints perfectly at half scale, and why "it always fails around the same height" is a chamber observation rather than a mechanical one.
What actually fixes it
Warm the chamber and keep it warm. An enclosure is not primarily about draughts; it is about the air the part is standing in. Even a cardboard box over an open-frame machine raises chamber temperature enough to change the outcome on ABS. A machine with an actively heated chamber effectively removes the failure.
Turn the part-cooling fan off for ABS, ASA, polycarbonate and PC-ABS. This is the change that feels most wrong to anyone whose instincts were built on PLA, and it is the single largest lever here. Those materials do not need cooling to hold an overhang the way PLA does, and every cubic centimetre of air moved across the part is weld strength removed.
Raise the nozzle by 10 °C and reprint the same file. More heat carried into each bead buys more time above the welding threshold.
Add walls rather than infill. More perimeters means more hot plastic per layer, which raises the local temperature and thickens the section that has to resist the tension.
Dry the filament if it is polycarbonate, a polyamide or a copolyester. Hydrolysed polymer welds badly no matter how warm the chamber is.
The ventilation point that belongs here
Advice to enclose a printer usually stops before mentioning what an enclosure does to the air in the room. ABS and ASA emit styrene and fine particulate while printing, and a sealed box concentrates them and then releases them when the door is opened.
Enclose the machine, and put it somewhere ventilated — a room you are not sitting in for hours, ideally with the enclosure's exhaust filtered or ducted outside. This is a genuine consideration rather than a legal footnote, and it applies to exactly the materials this page is about.
What makes it worse
Printing in a cold garage in winter, with an open frame, next to a door. Also: reducing print speed on the theory that slower is gentler. Slower means each layer takes longer, which means the previous one is colder when the next arrives — on this fault, slowing down is actively harmful for tall thin sections. Print those hotter, or fill the plate with copies so the toolhead has somewhere else to be between passes, but do not simply reduce speed and expect a stronger weld.
Which polymers split, and which never do
ABS and ASA above all, then polycarbonate and PC-ABS, then the polyamides, which crack for the same reason and add moisture sensitivity on top. PETG rarely delaminates in the printer — it welds well and contracts modestly. PLA almost never does, which is why anyone meeting this fault for the first time has usually just changed material rather than changed anything about their machine.