Reading a failed print
A failed print isn't a mystery — it's a record. Every defect on that part is a frozen snapshot of the moment things went wrong, and once you learn to read them, the part tells you what happened, where, and usually why. The stringy web across a set of towers, the corner that curled off the bed, the gap that opened in a top surface: none of these are random. Each is the signature of one specific failure in the chain that runs from your model to the nozzle to the plate. This article is the map. It teaches you to look at a failed part as evidence, points you to the defect you're seeing, and sends you to the article that explains its cause and its fix.
Look before you tweak
The instinct after a failed print is to change a setting and try again. Resist it for sixty seconds. A print that fails has already paid for the information — the least you can do is collect it before you bury it under a new attempt. Pick the part up, turn it in the light, and ask three questions in order: where on the part did it go wrong, when in the print did it happen, and what does the defect physically look like. Those three answers almost always converge on a single cause.
Where matters because FDM builds bottom to top, so height is time. A defect only at the base points at the first layer, the bed, or adhesion. A defect that starts partway up and continues to the top points at something steady — temperature, flow, cooling — that was wrong the whole time. A defect that appears once, at one height, points at a single event: a layer that had to bridge, an overhang that began, a travel move over open air.
The symptom-to-cause map
Most FDM failures fall into a handful of families. Find the row that matches what you're holding, and follow it to the article that covers the cause and the fix in depth. The design fix — the part of the problem you own in the model — always comes first there; the slicer and machine fixes follow.
| What you see | Most likely cause | Go to |
|---|---|---|
| Corner lifted off the bed; part detached mid-print | Warping, weak first-layer adhesion | Bed adhesion and warping |
| Fat, bulging base; holes tight near the plate | Elephant's foot, over-squished first layer | Bed adhesion and warping |
| Part snapped clean along a layer line | Weak layer bonding | Temperature and layer bonding |
| Visible cracks between layers, no impact | Delamination — nozzle too cold or over-cooled | Temperature and layer bonding |
| Fine hairs and webs strung between features | Stringing, oozing | Stringing and blobs |
| Bumps or scars at one point on each layer | Blobs, zits, seam | Stringing and blobs |
| Gaps between adjacent lines; thin, sparse walls | Under-extrusion | Extrusion problems |
| Rough, over-full surface; blobby top | Over-extrusion | Extrusion problems |
| Holes in the top surface; pillowing | Too few top layers, under-extrusion | Extrusion problems |
| Echoes or ripples beside sharp corners | Ringing, ghosting (vibration) | Dimensional and surface defects |
| Part measures off nominal; won't fit | Contraction, first-layer squish | Dimensional and surface defects |
| Drooping, curled overhang underside | Overhang sag, insufficient cooling | Overhang and bridge failures |
| Sagging strands across a gap | Failed bridge | Overhang and bridge failures |
When two families overlap
Some symptoms have more than one cause, and the trick is to use where and when to break the tie. A part that measures too small could be contraction (grows with part size, affects the whole part) or first-layer squish (only near the plate, only in the first few tenths). A rough surface could be over-extrusion (everywhere, the whole print) or stringing that got dragged and flattened (only near features you travel between). Delamination and weak layers look similar, but delamination shows visible gaps you can see into, while a weak bond looks fused yet snaps under load. The map above points you to the most common cause; the article it sends you to has the tiebreakers.
Where design ends and the machine begins
Throughout this section, every defect is split the same way: the part of it you fix in the model, and the part you fix at the machine. This matters because the two are not interchangeable. No retraction setting saves an overhang you should have reoriented; no reorientation fixes a nozzle that's genuinely too cold. Reading a failed print well means separating "I designed this to fail" from "the machine let me down" — and most failures, honestly, are a little of both. The articles that follow always give you the design lever first, because that's the one you own here in Kapy, and the one that travels with the part to any printer you send it to. If you want the mental model underneath all of it — why molten plastic behaves this way at all — start with How FDM Shapes Your Design.