2 Simple Methods to Anneal 3D Printed PLA
Annealing raises the heat deflection temperature of PLA parts, but it also shrinks them and can warp thin walls. This page covers two methods you can run with shop equipment, the temperatures and dwell times that work, and the part geometries where annealing does more harm than good.

What Annealing Actually Does to PLA
PLA is semi-crystalline at best. Heat lets the chains relax and pack, which raises the softening point but costs you dimensional accuracy.
Why FDM PLA Softens Early
As-printed PLA is mostly amorphous. The extruder lays down molten polymer, it cools fast, and the molecular chains freeze wherever they happened to land. That frozen disorder is why a printed bracket can start to sag at 55–60 °C, well below the 170–180 °C melting range you see on a filament datasheet. The material is not melting; it is passing its glass transition temperature, Tg, which for standard PLA sits near 60 °C.
Annealing holds the part above Tg for a set time, then cools it slowly. The chains get enough mobility to slide into more ordered regions, so the crystalline fraction rises. A higher crystalline fraction means a higher heat deflection temperature, typically 100–120 °C after a good anneal. It also means the part becomes stiffer and more brittle, and it will shrink 1–3% as the chains pack tighter.
That shrinkage is the trade you are making. Every dimension on the part moves, and the move is not uniform in all directions. XY dimensions usually shrink more than Z, because the layer bonds and the raster orientation give the Z axis a different constraint path. Plan for it in the model, or accept that holes and press fits will change size.
Method 1: Oven Annealing in a Kitchen or Lab Oven
A convection oven with a stable setpoint is enough. Ramp from room temperature to 60–70 °C, hold for 30 minutes so the whole part reaches temperature, then step up to 90–110 °C and hold for 1–4 hours depending on wall thickness. A 3 mm wall needs roughly 1 hour; a 10 mm solid block needs closer to 4. Cool inside the oven with the door closed, at no more than 1 °C per minute down to 40 °C.
The ramp and the cooldown matter more than the peak temperature. Skipping the slow cool re-quenches the part and puts most of the crystallinity back where it started. Pulling a hot part out onto a bench also makes it warp as the outside skins cool faster than the core.
Support the part during the whole cycle. PLA is soft above 60 °C, so a flat panel will sag under its own weight if it is only held at the edges. Fine sand, ceramic powder, or a machined fixture all work as a support. Do not use a metal tray in direct contact with the part if you can avoid it; the tray heats faster than the plastic and leaves marks.
Oven choice is a safety question as much as a quality one. A dedicated lab oven with a PID controller holds ±2 °C, which is good enough. A kitchen oven with a 25 °C swing will overshoot and slump thin sections. Measure the actual air temperature with a thermocouple, not the dial.
Method 2: Salt or Sand Packing
Packing the part in a bed of fine salt or sand does two jobs at once. The media supports the walls against gravity, and it slows heat transfer so the part sees a gentler ramp than it would in open air. This is the method to use for tall, thin, or hollow parts where oven annealing alone would collapse them.
Use dry, fine-grain salt or clean kiln sand. Fill the container in layers and tap it down so there are no voids, and bury the part at least 20 mm below the surface on all sides. The part itself goes in at room temperature, not preheated. Bring the whole bed up to 100–110 °C and hold for 1–3 hours, then let the container cool on the bench overnight.
Salt wins on support, but it has a cost. Salt is mildly abrasive and slightly hygroscopic, so rinse the part in warm water and dry it after the cycle. Sand leaves grit in small holes and threads. Neither media gives you a clean surface on fine detail, so mask or plug any feature you care about.
Media packing also limits how fast you can run the cycle. A 5 L container of salt takes 40–60 minutes just to reach setpoint, and longer to cool. That is fine for a batch of small parts packed together. For a single thin part, the oven method with a fixture is usually faster and easier to control.
Annealing Parameters by Part Type
Starting points only. Confirm with a test part before you run a full batch.
| Part type | Method | Hold temp and time | Expected shrinkage |
|---|---|---|---|
| Flat panel, 2–3 mm wall | Oven, sand support | 100 °C, 1 hour | 1–2% in XY |
| Bracket, 3–5 mm wall | Oven, fixture | 100 °C, 1.5 hours | 1–2% in XY, 0.5% Z |
| Tall hollow shell | Salt packing | 105 °C, 2 hours | 2–3% in XY |
| Solid block, 10 mm+ | Salt packing | 110 °C, 3–4 hours | 1–3%, near-isotropic |
| Small detail part | Oven, no support | 90 °C, 45 minutes | 1–2%, check thin ribs |
| Threaded insert boss | Oven, fixture | 95 °C, 1 hour | Threads tighten, re-tap |
When Annealing Is the Wrong Call
Skip it when the part is a fit check, a jig, or anything that has to mate with a machined counterface. A 2% shrink on a 50 mm bolt pattern moves each hole 0.5 mm, which is past most clearance allowances. Re-printing is cheaper than reworking a warped bracket.
Skip it when the geometry is mostly thin ribs, unsupported overhangs, or long slender pins. Those features are the first to sag and the last to hold tolerance, and no amount of salt packing fixes a 1 mm pin that bows.
Skip it when the service temperature is below 50 °C. A desk organizer or a display model gains nothing from the extra stiffness and loses surface finish to the media bed.
Use it when the part sees heat in service: an under-hood clip, a fixture near a heated bed, a duct that carries warm air. Those are the cases where the 30–50 °C gain in deflection temperature pays for the shrinkage and the cycle time.
If the part needs both heat resistance and tight tolerance, the honest answer is often a different process. Machined PEEK, POM, or a filled nylon will hold ±0.05 mm and take 150 °C without a shrink allowance. At GreatLight we run both routes, so we can tell you which one is cheaper for your geometry rather than pushing one over the other.
Annealing PLA: Common Questions
How much does PLA shrink when annealed?
Expect 1–3% linear shrinkage, with XY usually higher than Z. The exact number depends on wall thickness, infill, and how much crystallinity the filament can reach.
Print a test coupon of the same geometry and measure it before and after. That gives you a shrink factor you can apply in the slicer for the production parts.
Can I anneal PLA in a home oven?
Yes, but the temperature swing on a kitchen oven is often 20–30 °C, which is enough to slump thin walls. Use an oven thermometer or a thermocouple to check the real air temperature.
A better option is a small toaster oven with a PID controller, or a lab oven if one is available. Stable setpoint matters more than peak temperature for repeatable results.
Does annealing make PLA stronger?
It makes PLA stiffer and more heat resistant, not tougher. The gain is in heat deflection temperature, typically from about 60 °C to 100–120 °C.
Impact strength usually drops. The same ordering that raises the softening point makes the material more brittle, so parts that see shock loads may crack where they previously bent.
Why did my part warp during annealing?
The most common causes are a ramp that is too fast, a cooldown that is too fast, or a part that had no support above Tg. PLA is soft above 60 °C and will sag under its own weight.
Slow the ramp to about 1–2 °C per minute, cool inside the oven, and support the part in sand or a fixture for the whole cycle.
Can I anneal assembled or multi-material parts?
Not safely. Glue joints, threaded inserts, and materials with different shrink rates will move at different times, and the joint usually fails.
Anneal individual PLA components first, then assemble. If an insert must go in after annealing, heat-set it at a lower temperature and accept a local loss of crystallinity.
Is annealing worth it if the part needs tight tolerance?
Rarely. A 2% shrink on a 50 mm feature is about 1 mm of movement, and the shrink is not uniform across the part. That is hard to compensate in the slicer.
For tight tolerance plus heat resistance, a machined plastic such as POM, PEEK, or filled nylon is usually the better route. We can quote both and compare.
Need a Heat-Resistant Part, Not Just an Annealed One?
Send us the drawing and the service temperature. We will tell you whether annealing PLA is enough or whether a machined plastic is the cheaper answer.
12-hour quoteFree DFM analysis100% inspection