You take a part off the plate, it is perfect, you put it on the bench. In the morning there is a crack through it — sometimes with a faint ticking audible in the hours between, as anyone who has left a large ABS print in a quiet workshop can confirm.
Nothing touched it. No load was applied. The part failed on its own, which makes this the most unsettling failure on this site and one of the more interesting ones.
Stress that was frozen into the part
Every layer was laid down hot and contracted as it cooled, against layers below that had already finished contracting. Most of that tension is carried by the welds between layers while the print is running.
When the print finishes, the part is a body full of locked-in internal stress in equilibrium with itself. Polymer at room temperature is not perfectly rigid: chains creep slowly, the stress redistributes, and any place where a local concentration exceeds the local strength becomes a crack. That process takes hours because creep is slow, which is exactly why the failure is not immediate.
The stress is highest where the part is thickest and where the geometry changes abruptly — sharp internal corners, the junction between a thin wall and a solid boss, the base of a tall feature.
And, for some materials, crystallisation
Semi-crystalline polymers — the polyamides, polypropylene, PET, and to a lesser degree PLA — continue to organise their chains into crystalline regions after printing. Chains that have packed into an ordered region occupy less space than chains left in a tangle, so the part quietly loses volume while it sits there, over hours to days.
If different regions crystallise at different rates, which they will if the part cooled unevenly, the resulting differential shrinkage adds another stress field to the one already there. This is why a nylon part can crack on a shelf, and why nylon parts also change dimension for a week after printing.
Whether a polymer can pack its chains into ordered regions at all is decided by the shape of the backbone rather than by anything the printer did — which is why this failure sorts materials so cleanly, and why no profile change moves a material from one group to the other. Our sister site PeriodicDeck covers that side of the chemistry.
Releasing the stress on your terms
Cool slowly and evenly. On an enclosed machine, leave the door shut and let the whole chamber come down with the part in it. Taking a polycarbonate print out of a warm enclosure into a cold workshop is a thermal shock applied to a body that is already under stress, and it produces this failure reliably.
Reduce infill on thick parts. Counter-intuitive, and correct: more material inside means more contracting mass and more stored stress. Increase the wall count instead if strength is the aim — walls carry load better and add less internal tension.
Add fillets to internal corners. A sharp internal corner is a stress concentrator, and it is where the crack will start. A radius spreads what a corner concentrates, and on a printed part it costs nothing.
Keep the chamber warm during the print. A part built in a warm environment has less locked-in stress to release later, because each layer was less thoroughly chilled before the next arrived.
Anneal deliberately, or not at all. Heating a part slowly to just below its softening point and cooling it slowly relieves stress and, in semi-crystalline materials, completes the crystallisation on your schedule rather than the part's. It also distorts the part while doing it, so it needs a support jig and a measured shrinkage allowance. Done properly it is a process; done casually it is a second way to ruin the part.
Telling it apart from its neighbours
A crack that appeared during the print, running along a layer line, is cracking and delamination — a weld that lost to contraction while printing.
A part that broke under load along a layer line has weak layer adhesion.
A part that cracked with no load, hours later, often across layers rather than along one, is this page. The direction of the crack is the tell: interlayer failures follow the layer lines because that is where the weakness is, while a stress-relief crack goes where the stress is highest and will happily cut across beads to get there.
Where it happens most
Polycarbonate is the classic case, and thick polycarbonate parts removed from a hot chamber are the classic circumstance. The polyamides crack for the crystallisation reason as well as the stress one. Polypropylene shrinks more than anything else here and does it slowly. ABS cracks when large and thick. PET has the same tendency as its copolyester relatives but without their amorphous forgiveness.
PLA is largely exempt — it crystallises slowly enough at room temperature that most parts never notice — and PETG is deliberately formulated to stay amorphous, which is precisely why it is the easy functional material and why it does not do this.