How much does ABS shrink? XY, Z and the compensation value

ABS (acrylonitrile butadiene styrene) shrinks about 0.4 to 0.8% in X and Y as it cools, which takes 0.4 to 0.8 mm off a 100 mm part. Height is different: in one printed-part study, ABS came out 0.92% taller than the model, so the Z correction runs the other way.

How much does ABS shrink?

Printed ABS shrinks about 0.6 to 0.7% in X and Y. A Binghamton University study measured 0.59% in X and 0.69% in Y. Sunlu lists 0.4 to 0.8%, and molded ABS test bars shrink 0.4 to 0.7%. Prusa quotes a wider 1 to 2%. In the same study, ABS grew 0.92% in Z.

Source	What was measured	X and Y	Z
Weiser et al., Binghamton (2021)	Printed parts on a QIDI X-One2, digital calipers	0.59% (X), 0.69% (Y)	-0.92% (grew)
Sunlu performance table	ABS filament, method not stated	0.4 to 0.8%	not given
Spectrum Smart ABS data sheet	Molded 3.2 mm bar, ASTM D955	0.4 to 0.7%	not given
INEOS Styrolution Terluran GP-35	ABS resin, molded, ISO 294-4	0.4 to 0.7%	not given
Prusa Knowledge Base, ABS	General guidance, method not stated	“around 1-2%”	not given

Sunlu also lists 0.3 to 0.7% for its transparent ABS, 0.8% for its high-speed ABS and 0.5% for ABS-GF.

Bambu Lab’s ABS sheet (V3.0) and Polymaker’s PolyLite ABS sheet (V5.6) give no shrinkage figure. eSUN’s ABS Odorless sheet describes shrinkage in words only.

For a starting value, use the Binghamton XY average of 0.64%. It sits inside both the molded range and Sunlu’s. Bambu Lab’s wiki notes that different part shapes show different errors, so confirm the figure with a test part (steps below).

Why do ABS prints grow in Z?

The Binghamton study measured ABS parts 0.92% taller than designed and offered no explanation. A printed part’s height is built by the nozzle stepping up one layer at a time, so first-layer height and where you measure affect the Z reading as well as the material. Measure your own part before trusting any Z value.

What the sources do say:

• It was ABS only. In the same study, PLA (polylactic acid) shrank 0.80% in Z and PETG (glycol-modified polyethylene terephthalate) 0.73%. Only ABS grew.
• Reheated prints grow in Z too. When Prusa annealed printed PLA, PETG, ASA (acrylonitrile styrene acrylate) and ABS samples, all of them shrank most along their length and “extended” in Z. D’Amico et al. (2017) found that reheated ABS strains mainly along the printed lines. They attribute this to residual stress from rapid cooling during printing.
• The first layer counts. If the first layer goes down thicker than the sliced layer height, the whole part reads taller.

Measure Z from the bed face to the top, with no brim or raft. Positive shrinkage means it came out short. Negative means it grew. Enter that value as is.

Does ABS shrink more than PLA and PETG?

ABS shrinks more than PLA in most published figures and about as much as PETG. Printed parts in the Binghamton study shrank 0.64% in X and Y for ABS, 0.65% for PETG and 0.49% for PLA. Sunlu lists ABS at 0.4 to 0.8%, PETG at 0.1 to 0.5% and PLA at 0.1 to 0.3%.

Material	Printed X	Printed Y	Printed Z	Sunlu	Molded resin
PLA	0.63%	0.35%	0.80%	0.1 to 0.3%	0.4% (NatureWorks Ingeo 3052D)
PETG	0.76%	0.54%	0.73%	0.1 to 0.5%	none checked
ABS	0.59%	0.69%	-0.92%	0.4 to 0.8%	0.4 to 0.7% (Terluran GP-35)

Printed columns are from the Binghamton study (same printer and calipers for all three).

Two things follow:

• ABS against PLA: ABS shrinks as much or more in every source here. On a 100 mm printed part the Binghamton gap is 0.15 mm in XY.
• ABS against PETG: they are level in printed XY, 0.64% against 0.65%. Prusa, though, says PETG “does not shrink or warp” and lists significant warping as ABS’s main drawback.

The Z column matters most when you switch filaments. PLA and PETG came out short in Z and ABS came out tall. A PLA profile’s Z value copied to ABS makes the height error worse, not smaller.

ASA, its outdoor relative, sits in the same molded band: INEOS lists 0.4 to 0.7% for Luran S 777K. The ABS vs ASA shrinkage page compares every published figure.

Why ABS warps when it only shrinks 0.6%

Warping is shrinkage the bed holds back. The first layer stays stuck to a hot plate while the layers above cool and contract, so the stress pulls the corners up. Prusa says corner lifting happens “mostly” with high-temperature materials such as ABS. It blames the sudden drop from nozzle temperature to ambient air.

The fixes reduce that drop rather than the shrinkage:

• An enclosure. Prusa’s ABS guide asks for a bed of at least 100 °C, and Bambu Lab’s ABS sheet gives a 45 to 60 °C chamber. The chamber heater calculator sizes a heater for your enclosure.
• A draft shield (PrusaSlicer, Skirt and brim), and no draft from room ventilation.

Shrinkage compensation does not fix warping. A warped part also measures wrong, so stop the warping before you measure shrinkage.

How to calculate ABS shrinkage from a test print

1. Set the filament’s shrinkage to 100% in OrcaSlicer (0% in PrusaSlicer) and keep the model at 100% scale.
2. Print a block about 100 mm long and 20 mm tall in your ABS, with your real enclosure, bed temperature and fan settings.
3. Let it cool to room temperature. Measure X and Y mid-face and Z without the brim.
4. For each axis: shrinkage = (designed - measured) / designed.

Say a 100 mm length in X measures 99.36 mm: (100 - 99.36) / 100 = 0.64%. A 20 mm height that measures 20.18 mm gives (20 - 20.18) / 20 = -0.9%, a part that grew.

The shrinkage compensation calculator takes the designed and measured sizes and returns every slicer’s value. To separate shrinkage from flow and hole errors, see how to calibrate dimensional accuracy per material.

ABS shrinkage compensation in OrcaSlicer

OrcaSlicer has two filament fields, Shrinkage (XY) and Shrinkage (Z), visible in Advanced mode. Each takes the cooled size as a percentage of the design, so 100% means no change. The source accepts 50 to 150%. The base profile in OrcaSlicer’s filament library sets 100%, and its base ABS profile keeps that default, so any ABS value has to come from you.

With the Binghamton ABS figures (0.64% in XY, -0.92% in Z):

Where	Value to enter	Effective model scale
OrcaSlicer Shrinkage (XY)	99.36%	100.64%
OrcaSlicer Shrinkage (Z)	100.92%	99.09%
Bambu Studio Shrinkage (XY only)	99.36%	100.64%; scale Z to 99.09% by hand
PrusaSlicer Shrinkage compensation XY / Z	0.64% / -0.92%	100.64% / 99.09%

OrcaSlicer scales by 100 divided by the value, so 100 / 99.36 = 1.0064 in XY and 100 / 100.92 = 0.9909 in Z. Z goes above 100% because the part grew. If your test part shrank in Z instead, the value drops below 100%.

Use one method only. A value in the filament profile plus a scaled model compensates twice.