How to print dimensionally accurate nylon parts

To print dimensionally accurate nylon parts, dry the filament and keep it below 20% relative humidity (RH) while printing, and print in a closed, warm chamber. Then compensate for two separate changes: shrinkage as the part cools and crystallizes, and swelling as it absorbs water afterwards. Decide whether the size that matters is the dry part or the conditioned one, measure a test part in that state, and scale from that measurement.

Why nylon is hard to hold to size

PLA and PETG mostly change size once, while they cool. Nylon changes size three ways: it contracts and crystallizes as it cools, it can keep shrinking when heated later, and it grows as it absorbs water.

Water absorption and swelling

Polyamides (PA, the chemical family name for nylon) absorb water from the air and grow as they do. BASF’s Ultramid brochure puts a number on it: with the water evenly spread through the part, PA6 and PA66 show “a mean increase in length of 0.2 to 0.3 % per weight percent of absorbed water”. Glass-fiber grades change less than 0.1% along the fibers.

How much water a part takes up depends on the polyamide type. Water bonds to the amide groups in the chain, and PA6 has more of them than PA12. Evonik notes that PA12 has the lowest concentration of amide groups of any commercial polyamide, and that its parts show “almost no dimensional changes with variations in humidity”.

Material	Water at saturation, 23 °C	Water at 23 °C, 50% RH	Source
PA6, unfilled (Ultramid B3S)	9.5%	3%	BASF data sheet
PA6, 30% glass fiber (Ultramid B3EG6)	6.6%	2.1%	BASF data sheet
PA12, unfilled (VESTAMID L grades)	1.3 to 1.6%	0.7 to 0.8%	Evonik brochure

Filament makers publish moisture uptake on a different basis. Bambu Lab lists a “saturated water absorption rate” at 25 °C and 55% RH:

Filament	Base polymer	Water absorption, 25 °C, 55% RH
Bambu PA6-GF	Nylon 6, glass fiber	2.56%
Bambu PA6-CF	Nylon 6, carbon fiber	2.35%
Bambu PAHT-CF	PA12 and other long-chain PA, carbon fiber	0.88%

Multiply the two figures to size the swelling. A PA6 part that picks up 3% water at 23 °C and 50% RH grows by roughly 0.6 to 0.9% in length (3 × 0.2 to 3 × 0.3). On a 100 mm part that is 0.6 to 0.9 mm, as large as the print shrinkage you set out to correct.

Crystallinity and shrinkage

Nylon is semi-crystalline. As it cools, part of it packs into ordered crystals, which takes up less volume than the melt. BASF lists the causes of molding shrinkage as cooling, the “change in the state of aggregation” and crystallization. Amorphous plastics such as ABS and PETG have no crystallization step.

Resin data sheets give the shrinkage and the coefficient of thermal expansion (CTE), and show what fibers do to both. These are injection-molding figures, not printed-part figures, but they show how the polymer families differ:

Material	Molding shrinkage	Thermal expansion (CTE)
PA6 (Ultramid B3S)	0.9% along and across flow	102 × 10⁻⁶ /K along flow
PA6-GF30 (Ultramid B3EG6)	0.2% along flow, 0.7% across	22 × 10⁻⁶ /K along, 106 × 10⁻⁶ /K across
PA12 (VESTAMID L, unfilled)	not compared here	140 to 150 × 10⁻⁶ /K, 23 to 55 °C

The glass-fiber row is the useful one for filled filaments. The fibers hold their own direction almost still, while the cross direction behaves close to the unfilled polymer. Printed fibers tend to follow the extrusion path, so a fiber-filled part can shrink differently along a perimeter than across it. Measure X and Y separately.

Warping from residual stress

Each new line of nylon goes down hot onto a cooler layer and shrinks as it cools, and the layer below resists. The stress that builds up pulls corners off the bed and bows long, flat parts. Prusa’s nylon guide calls warping “the main disadvantage of pure polyamide”. It notes that carbon-fiber filled nylons warp far less and that a high ambient temperature from an enclosure “helps a lot”.

Stress that stays in the part can also move its dimensions later. BASF describes post-shrinkage: molded parts can change size over time as internal stresses and orientations relax and further crystallization takes place. The change is small at room temperature but can be significant at higher temperatures, and annealing speeds it up.

Dry nylon filament before and during printing

Water boiling in the nozzle leaves voids and a rough surface, and both ruin a caliper reading. Every maker below sets a drying step, and the carbon- and glass-fiber sheets add a humidity limit for the spool while it prints.

Filament	Drying before printing	Humidity while printing and in storage
Bambu PA6-CF, PA6-GF, PAHT-CF	Forced-air (blast) drying oven, 80 °C, 8 to 12 h	Below 20% RH, sealed with desiccant
Polymaker PolyMide CoPA	100 °C, 8 h	Below 20% RH
Prusa guidance for polyamides	At least 4 h below 90 °C	Stored airtight with desiccant

The humidity limit matters as much as the oven: a sealed dry box keeps a long print from re-absorbing water halfway through. Bambu Lab’s data sheets also note that its spools are rated to 90 °C, and that a microwave or kitchen oven can damage filament because it heats unevenly.

For a symptom check and a way to weigh how much water a spool has taken up, use the wet filament checklist. Its weighing method works best on nylon, because the numbers are large: a 1 kg spool of PA6-CF at 2.35% holds about 23.5 g of water.

Chamber, bed and cooling settings for nylon

Slow cooling is the main defense against warping, and the data sheets agree on that. Their numbers differ because each formulation behaves differently.

Setting	Bambu PA6-CF	Bambu PAHT-CF	Bambu PA6-GF	Polymaker PolyMide CoPA
Chamber	45 to 60 °C	45 to 60 °C	45 to 60 °C	Enclosure needed, ambient temperature
Bed temperature	80 to 100 °C	80 to 100 °C	80 to 100 °C	25 to 50 °C
Build surface	Engineering, High Temperature or Textured PEI plate	Same	Same	PC or textured PEI
Surface preparation	Glue	Glue	Glue	Glue when needed
Part cooling fan	0 to 60%	0 to 40%	0 to 60%	Off

Prusa’s guide lists its PA Nylon sheet as the surface with the best adhesion for most polyamides. The other makers list textured polyetherimide (PEI) or dedicated plates with glue.

The low-bed CoPA figure is not an error. Polymaker’s annealing page says its nylons are made to crystallize slowly so they warp less, and that they do not reach full crystallization without annealing. That trade matters for dimensions: a part from such a filament is expected to change a little when you anneal it.

If your printer has no active chamber heat, the chamber heater calculator estimates the heater wattage an enclosure needs to hold 45 to 60 °C from its size, panels and insulation.

What annealing does to nylon dimensions

Annealing heats the part below its melting point so the polymer can finish crystallizing and residual stress can relax. The makers give these schedules:

Polymaker nylons: 80 to 100 °C for 6 to 16 hours, “to fully crystallize”.
Bambu PA6-CF, PA6-GF, PAHT-CF: 80 to 130 °C for 6 to 12 hours. Bambu Lab adds that some prints may deform and warp after annealing.
BASF, for thick molded parts: 12 to 24 hours at 140 to 170 °C, which gives “slight post shrinkage and sometimes a small amount of warpage”.

Every source says the same thing about size: annealed nylon comes out slightly smaller. None of them publishes a percentage for printed parts. Polymaker says it cannot give a standard figure, because the change depends on geometry, size and infill.

Annealing also dries the part, and Polymaker notes that the nylon “will slowly moisture condition after”. In its tests on 40 mm cubes, most nylons shrank slightly on annealing and then grew past their original size after moisture conditioning. PA12-CF was the exception, with little growth. So the order of operations is fixed: print, anneal, then condition, and measure after each step.

Dry-as-printed vs conditioned: when to measure

A nylon part has no single size. It is smallest right after annealing and grows toward an equilibrium set by the humidity around it, so pick the state the drawing refers to.

Dry-as-printed. Right off the printer or out of the annealing oven. This is the size most people measure, and the size nylon will not keep in normal indoor air.
Conditioned. Moisture content in balance with the service air. BASF’s data sheets list each property in a dry and a conditioned column. The B3S sheet gives a tensile modulus of 3,500 MPa dry and 1,200 MPa conditioned.

BASF writes that Ultramid parts “only achieve their optimum impact strength and largely constant dimensions after absorbing moisture”. For a part that must fit in normal indoor air, the conditioned size is the one to design to.

ISO 1110 conditioning

The plastics industry has a standard shortcut for this. ISO 1110 describes accelerated conditioning of polyamide test specimens at 70 °C and 62% relative humidity, held until the specimen reaches at least 95% of its equilibrium moisture. The equilibrium it reaches is close to that of the standard atmosphere of 23 °C and 50% RH. BASF recommends this climate chamber method over a hot water bath, which can cause stains, deposits and warping in thin parts with internal stresses.

Conditioning time rises sharply with wall thickness. BASF describes the dependence as quadratic, so a wall twice as thick takes about four times as long.

Without a climate chamber, Polymaker’s moisture conditioning page gives home methods: 48 hours in a humid container or submerged in water, or about two weeks in ambient air, always after annealing. These do not reproduce ISO 1110, but they bring a test part near the state it will reach in use.

Measure at a known temperature

Nylon also has a high CTE. With 102 × 10⁻⁶ /K for PA6, a 100 mm part reads 0.05 mm longer at 28 °C than at 23 °C. With PA12’s 150 × 10⁻⁶ /K the same 5 °C gap is 0.075 mm. Let the part and the calipers sit at room temperature before you measure, and note that temperature.

Design allowances: scaling, holes and press fits

Scaling for shrinkage and swelling

Shrinkage and swelling are both proportional to size, so a percentage scale corrects them, one number per axis:

print shrinkage      s1 = (designed - measured dry) / designed
moisture growth      g  = (measured conditioned - measured dry) / measured dry
net size ratio       r  = (1 - s1) x (1 + g)
scale factor         k  = 1 / r

OrcaSlicer’s filament profile takes the ratio r directly as its Shrinkage (XY) and Shrinkage (Z) values: “94% if you measure 94mm instead of 100mm”. Use either the filament field or the model scale, not both. The shrinkage calculator turns a measured part into the value for each slicer. Its annealing fields accept a negative number for growth, so it can apply the moisture step too.

Hole compensation

Scaling barely moves a small hole. Holes print undersized partly from the flat facets of an exported circle and partly from extrusion width, and a fixed offset corrects both better than a percentage. OrcaSlicer’s X-Y hole compensation grows or shrinks holes by a set distance, and “positive values make holes bigger”. X-Y contour compensation does the same for outer walls, and polyholes print round holes as polygons whose flats land on the intended diameter. For a full calibration routine, see the guide to dimensional accuracy calibration.

Press fits and clearances against metal

Swelling works against fits with parts that do not swell. With water spread evenly, a nylon part grows as a whole, and a hole in it grows too. At the 0.75% growth used below, an 8.00 mm hole for a steel dowel opens by 0.06 mm as the part conditions. A press fit tuned on a dry part can loosen, and a nylon pin in a metal bore can tighten.

No published clearance table covers printed nylon, so derive clearances from a test:

Print a coupon with a row of holes or bosses, stepping the size by 0.05 mm around the nominal.
Anneal it if the real part will be annealed, then condition it to the service state.
Try the mating part in each size and keep the one that gives the fit you want.
Put that offset in the CAD model or in hole compensation, not in the global scale.

A PA12-based or fiber-filled filament makes the whole problem smaller. With 0.7 to 0.8% moisture at 23 °C and 50% RH, unfilled PA12 has about a quarter of PA6’s 3%.

Worked example: a compensation factor for a conditioned PA6 part

A bracket in unfilled PA6 has a 100.00 mm dimension in X that must be right in normal indoor air. A test bar is printed with no shrinkage compensation, annealed, and measured dry at 23 °C: 99.10 mm. These readings are hypothetical. Use your own.

Print and annealing shrinkage. s1 = (100.00 − 99.10) / 100.00 = 0.90%.
Moisture growth. With no conditioned measurement yet, estimate it: 3% water at 23 °C and 50% RH (BASF’s B3S figure) times 0.25% per percent of water, the middle of BASF’s range, gives g = 0.75%.
Conditioned size of the test bar. 99.10 × 1.0075 = 99.84 mm.
Net ratio. r = 0.9910 × 1.0075 = 0.99843.
Scale factor. k = 1 / 0.99843 = 1.00157, so scale X to 100.16%, or enter 99.84% in OrcaSlicer’s Shrinkage (XY).

Compare that with correcting from the dry reading alone. The dry-only factor is 1 / 0.991 = 100.91%. It gives a perfect 100.00 mm part on the day it is annealed, and a 100.75 mm part once it has conditioned.

The estimate in step 2 carries a spread. Across BASF’s 0.2 to 0.3% range the conditioned bar lands between 99.70 and 99.99 mm, so the scale falls between 100.01% and 100.31%. That 0.3 mm window is why step 2 should be replaced by a measurement. Condition the test bar, measure it again at 23 °C, and enter the conditioned reading directly as the measured size in the shrinkage calculator.

Checklist for accurate nylon parts

Dry the spool on the maker’s schedule (80 °C for 8 to 12 h or 100 °C for 8 h in the sheets above), and print from a sealed box below 20% RH.
Print in a closed chamber at the data sheet’s temperature, 45 to 60 °C for the fiber-filled PA sheets above. Keep part cooling low or off as the sheet says.
Use the build surface and glue the data sheet lists, and keep the brim or first layer out of your Z measurement.
Decide whether the drawing means dry-as-printed or conditioned. For indoor service parts, it is usually conditioned.
Print a test bar about 100 mm long with compensation reset, and anneal it if the real part will be annealed.
Measure dry at a known room temperature, then condition it (ISO 1110 at 70 °C and 62% RH if available, or a humid container) and measure again.
Compute the net ratio (1 − s1) × (1 + g) for X, Y and Z separately. Fiber-filled parts can differ by direction.
Enter one correction: the slicer’s shrinkage field or the model scale, never both.
Correct holes with hole compensation or polyholes, and set press-fit clearances from a conditioned coupon.
Record the brand, color, drying, annealing and conditioning with the value, and re-measure when any of them change.