Annealing: when it adds strength, and when it just shrinks the part

Annealing is the one post-processing step that changes the material itself, not just its surface. It's also the most over-promised step online: videos of an annealed PLA part surviving boiling water rarely show how much the part shrank or how it warped.

Annealing does two different things

One: relieving internal stress

Every FDM part carries stress, because each layer is extruded hot and cools quickly on top of a layer that has already cooled. Holding the part warm enough for the chains to move a little lets that stress out. The result is a part less likely to crack unexpectedly and less likely to warp over time — it warps in the oven, not three months later on a customer's desk. This applies to every plastic.

Two: raising crystallinity

Only semi-crystalline plastics can do this. In PLA, PA, PP or PPS, chains can rearrange into ordered regions when held at the right temperature. Crystalline regions are stiffer, more heat-resistant and more solvent-resistant than the amorphous material around them, so as the crystalline fraction rises, modulus and heat deflection temperature rise with it. This is the part that changes mechanical properties, and it doesn't happen in PETG, ABS, ASA or PC — those are amorphous, and however long you anneal them you only get the first effect.

What each material gains

The price: the part will shrink, and it will warp

As crystalline regions form, chains pack tighter and the material takes up less volume. The part shrinks — usually in X and Y, and grows in Z, because the print orientation makes shrinkage anisotropic. The amount depends on shape, wall thickness, infill density and even infill pattern, so there's no general figure to look up: it has to be measured on a test piece printed with the same settings as the real part.

• —Thin-walled and tall parts warp the most. They don't have enough material to hold their shape once the plastic softens.
• —Hollow parts can collapse. Sparse infill means fewer internal columns.
• —Fibre-filled parts distort less. This is the most practical advantage of PLA-CF and PA-CF — the fibre resists chain movement, so the part holds its shape through annealing much better.
• —The surface resting on the tray will take its imprint. Soft plastic on a metal rack carries the rack's pattern away with it.

The process

1. 01

   Print two parts, not one

   One to anneal as a trial, one to compare. Same spool, same settings, same orientation. Without a test piece, every compensation figure after that is a guess.

2. 02

   Measure before annealing

   Measure the critical dimensions with calipers and write them down. After annealing, measure the same dimensions again. The difference is the compensation factor you'll put into the drawing for the next print.

3. 03

   Support the part while annealing

   Burying it in fine dry sand, dry salt or stone powder is a simple way to stop it sagging and sticking to the tray. For a flat part, a clean smooth flat surface is enough.

4. 04

   Start below the threshold, and step up

   Set the oven below the material's glass transition, hold, then step up in small increments on later trials. Going straight to the highest temperature you've heard of is the fastest way to get a distorted part.

5. 05

   Hold long enough for the part to heat through

   A thick part needs more time than a thin one — annealing time follows the thickest section, not the outer size.

6. 06

   Cool slowly in the oven

   Switch off, keep the door closed, and let the part cool with the oven. Taking a hot part out into room air puts back the very stress you just spent hours taking out.

The equivalent step for resin

Resin parts have a thermal step that does exactly this, and there the effect is published in numbers. Formlabs gives two columns for High Temp Resin: after UV cure, HDT at 0.45 MPa is 142 °C; after an additional thermal post-cure at 160 °C, it's 238 °C. Same part, same resin, one oven step apart. The wash and cure process is in washing and curing resin prints.

Annealing extracts the last of the properties a material already has. It doesn't give you properties the material doesn't have.

When not to anneal

1. 01When the part must fit precisely. A few percent of shrinkage is enough for a bearing seat to stop fitting. If you still need to anneal, anneal a test piece, measure, and scale the drawing by the measured factor.
2. 02When the material is amorphous and what you need is heat resistance. Annealed PETG still softens at the same temperature. You need a different material, not a different process.
3. 03When the part is already the right material for the job. If it's already ASA or PC and working well, annealing only adds distortion risk.
4. 04When you need a part for a car cabin. That's a material-selection problem, not a post-processing one — filament for car parts explains why that threshold rules PLA out even after annealing.

If what you really want is a stronger part rather than a more heat-resistant one, most of the gain lies in print settings, not the oven: wall count, nozzle temperature and orientation change more — see print settings for strength.