Annealing FDM parts means holding a printed component at a controlled temperature above its glass transition point long enough for the polymer to relax internal stress and, in semi-crystalline materials, grow more crystalline structure — which raises interlayer strength and heat deflection but almost always shrinks the part in the print plane. At Layer X, we anneal on request for functional FDM work, and the single question that decides whether it helps or ruins the part is always the same: can your design tolerate the dimensional change? This guide gives the time and temperature ranges we work to for each material, and the shrinkage behaviour you must compensate for in CAD before the part ever reaches the oven.
Why annealing works — and what it costs you
An FDM part is built line by line, layer by layer, and it cools unevenly. That leaves two problems locked into the geometry: frozen-in internal stress from rapid cooling, and weak bonds between layers where the deposited road never fully fused to the one below. Heat is the lever for both. Held above its glass transition temperature (Tg) but below its melting point, the polymer chains gain enough mobility to relax stress and diffuse across the layer boundaries, strengthening the Z direction that is normally the weakest axis of any FDM part.
In semi-crystalline polymers such as Nylon PA12, there is a second, larger prize: additional crystallisation. Printing quenches these materials so fast that they solidify only partly crystalline. A proper anneal lets crystals nucleate and grow, which is what delivers the big jump in stiffness and heat resistance. The cost is that crystals pack more densely than amorphous polymer, so the part contracts — and it does so anisotropically, more in some directions than others. Tensile behaviour before and after is best characterised to a standard such as ISO 527 for plastics, and heat resistance to ASTM D648 (heat deflection temperature), so you are comparing like with like rather than trusting a feel test.
The recipe table — by material
The numbers below are the published, well-established ranges we start from for each polymer we run on our FDM line — PLA, PETG, ABS, ASA and Nylon PA12. Treat them as starting points, not gospel: oven calibration, part mass and wall thickness all shift the real optimum, which is why we always qualify a recipe on a test coupon before committing a batch.
| Material | Approx. glass transition (Tg) | Typical anneal band | What you gain | Dimensional risk |
|---|---|---|---|---|
| PLA | ~55–60°C | ~80–110°C | Large stiffness and heat-resistance gain via crystallisation | High — warps and slumps readily without support |
| PETG | ~80°C | ~65–75°C (stress relief only) | Modest stress relief; little crystallisation | Moderate; low payoff, rarely worth it |
| ABS | ~105°C | ~90–100°C | Stress relief, fewer micro-cracks, steadier Z strength | Moderate; amorphous, so mainly relaxation not shrink |
| ASA | ~100–105°C | ~90–100°C | Stress relief while keeping UV/weather stability | Moderate; similar to ABS |
| Nylon PA12 | ~40–50°C | ~120–150°C | Largest strength and heat-deflection gain of any FDM polymer | High — must be dried first; significant XY contraction |
Two patterns fall out of that table. Amorphous materials — PETG, ABS, ASA — have little crystalline structure to grow, so annealing buys you stress relief and better layer fusion, not a dramatic property change, and the dimensional movement is smaller. Semi-crystalline materials — PLA and especially Nylon PA12 — are where annealing earns its keep on strength, and precisely where the shrinkage is largest. There is no free strength.
Time, ramp and cooldown — the part everyone skips
Temperature gets all the attention, but the schedule around it decides whether the part survives. We work to three principles:
- Ramp slowly. Placing a cold part into a hot oven sets up a thermal gradient that induces exactly the stress you are trying to remove. Bring the part up with the oven from cold where possible.
- Soak by mass, not by clock. A thin bracket equalises in well under an hour; a thick, dense block needs considerably longer for the core to reach temperature. Under-soaking anneals only the skin and leaves a stressed core.
- Cool slower than you heated. The single most common cause of a warped annealed part is pulling it out hot and letting it cool unevenly on a benchtop. Let it cool inside the oven, ideally supported.
For parts prone to slumping — tall, thin, or PLA in particular — we support the geometry during the soak, either bedded in a heat-stable medium or held in a simple fixture, so gravity does not do to the part at temperature what the print never intended.
The shrinkage you must design for
This is the heart of the matter and the reason annealing lives in a specialist post-processing guide rather than a general finishing article. Annealing is not dimensionally neutral. As a semi-crystalline part crystallises it contracts in the X and Y (print-plane) directions and frequently grows slightly in Z, because the layers relax and the material redistributes. The exact figures depend on material, infill, wall thickness and print orientation — which is exactly why we never quote a universal shrinkage percentage. Anyone who does is guessing.
Instead, we characterise it. Before annealing a production batch we print a test coupon in the same material, orientation and infill as the real part, measure it, anneal it on the intended recipe, and measure it again. That gives a per-axis scale factor we then apply back into the CAD model, so the part comes out of the oven on-size rather than the printer. It is the same discipline injection moulders use for mould shrink compensation, applied to additive.
Why it matters for fit: our standard FDM tolerance is ±0.3mm, and tighter interfaces are held to ±0.2mm on process-qualified parameters. A raw anneal can easily move a feature further than that entire tolerance band, so a bore that was a perfect press-fit before the oven becomes loose after it. If your part has mating features, bearing seats, threaded inserts or assembly interfaces, you cannot anneal it as an afterthought — the shrinkage has to be designed in from the start. Our wider notes on holding size are in the 3D printing tolerances and dimensional accuracy guide.
When to anneal — and when not to bother
Annealing is worth the extra step when:
- The part is load-bearing in the Z direction and layer separation is a real failure mode — brackets, clips, levers pulled across the layer lines.
- The part sees sustained heat near or above the printed material's untreated deflection point — enclosures near motors, under-bonnet fixtures, parts left in a hot vehicle.
- You are running Nylon PA12 and want its full mechanical potential, which the as-printed state never delivers.
It is usually not worth it when the part is cosmetic or dimensionally critical with tight mating features and no annealing compensation budgeted, when PETG is the material (low payoff for the risk), or when a smarter print strategy would solve the problem more cheaply. Often the honest answer to weak Z strength is not the oven at all but orientation: reorienting so the load runs along the layers rather than across them, covered in our guide to FDM design for production orientation, strength and layer adhesion. If a part is warping in the first place, fix that before you add heat — see warping causes and solutions, because annealing a warp-prone geometry simply anneals the warp in.
An FDM strength decision, not a blanket step
The practical workflow we recommend is a short decision chain. First, ask whether orientation or material choice removes the need to anneal at all — the cheapest anneal is the one you do not do. Second, if the strength or heat requirement genuinely needs it, pick the material band from the table above and lock a recipe on a coupon. Third, measure the coupon's per-axis shrinkage and compensate the CAD. Only then run the batch. Skipping the coupon step is how good parts come out of an oven the wrong size.
For parts where even an annealed FDM polymer cannot meet the strength or temperature target, the right move is a different process rather than a hotter oven — laser-sintered SLS Nylon PA12 arrives far more isotropic and dense straight off the bed, with none of the layer-adhesion penalty that annealing exists to patch. Choosing between them is a design decision we are happy to make with you.
Anneal with the fit designed in
Annealing is one of the highest-leverage post-processes in FDM when it is planned — and one of the fastest ways to scrap a batch when it is bolted on at the end. At Layer X we run it as an engineered step: qualified recipe, coupon-measured shrinkage, CAD compensation, controlled cooldown, all under our ISO 9001:2015 quality system from our Ahmedabad facility. If you have a functional FDM part that needs more Z strength or heat resistance, upload your CAD file for a 24-hour quote and tell us the loads it sees — we will advise whether to anneal, reorient, or move it to a different process before a single part is printed.