Carbon and glass fiber filament shrinkage: PETG-CF, PLA-CF, ABS-GF

PETG-CF (PETG filled with chopped carbon fiber) shrinks about 0.3 to 0.6%. That is the molding shrinkage 3D4Makers publishes for its 10% carbon fiber PETG, the same along and across the flow. Plain PETG printed parts shrank 0.54 to 0.76% in X and Y in a Binghamton University study, so start PETG-CF near 0.45% and measure your own part.

Do carbon fiber and glass fiber filaments shrink less?

Usually yes, mostly along the direction the fibers point. Bambu Lab’s wiki says the fibers limit the base polymer’s shrinkage. Polymaker’s glass-filled HT-PLA-GF shrinks 0.05% in X-Y against 0.88% for plain HT-PLA. Across the fibers the effect is weaker: BASF’s 30% glass-filled nylon 6 shrinks 0.2% along the flow but 0.7% across it.

The data has exceptions. Sunlu’s comparison table lists PLA-CF at 0.3 to 1.1%, higher than the 0.1 to 0.3% it gives plain PLA. Fiber helps most along the beads, so a part whose critical size runs across them can still shrink like the base plastic.

PETG-CF shrinkage rate: the published numbers

Only one PETG-CF figure turned up in the data sheets checked for this page. The nearest other data comes from plain PETG and a carbon fiber PET.

Material	What was measured	Shrinkage
3D4Makers PETG Carbon (10% CF)	Molding shrinkage, method not stated	0.30 to 0.60%, same along and across
Sunlu PETG-CF	Comparison table	No figure listed
Polymaker Fiberon PET-CF17 (PET, not PETG)	Printed 40 mm cube, 30% infill	0.15% in X-Y; grew 0.63% in Z
Plain PETG, Weiser et al. 2021	Printed parts, one printer	0.76% in X, 0.54% in Y, 0.73% in Z

The Fiberon percentages are arithmetic from the millimeters on Polymaker’s sheet: the cube measured 39.94 mm in X-Y and 40.25 mm in Z.

To turn the 0.45% midpoint into a setting:

scale = 1 ÷ (1 − 0.0045) = 1.00452, a 100.45% model scale
OrcaSlicer or Bambu Studio Shrinkage field: 99.55%
PrusaSlicer Shrinkage compensation XY: 0.45%

Bambu Studio’s own filament profiles apply no compensation: the shared base profile sets Shrinkage to 100%, and none of the Bambu Lab filament profiles changes it. The shrinkage compensation calculator turns any of these figures, or your own measurement, into each slicer’s value.

Tian et al. (2025) tested PETG-CF printing settings for dimensional accuracy and found heated bed temperature mattered most, ahead of infill density, layer height and nozzle temperature. Keep the bed temperature fixed between your test print and the real part.

PETG-CF after annealing

PETG-CF changes far less in the oven than plain PETG. Stojković et al. (Materials, 2023) annealed 165 mm tensile bars at 60 to 100 °C for 30 to 90 minutes:

Material	Average length change	Largest length change
PETG-CF	0.28 mm (0.17%)	1 mm (0.61%)
PETG	3.33 mm (2.0%)	22 mm (13%)
PLA	3.55 mm (2.2%)	5.8 mm (3.5%)

A filled grade can still move a lot at higher heat. Polymaker’s PET-CF17 cube lost 1.5% in X-Y after 10 hours at 120 °C, ten times its as-printed figure, and grew 0.83% in Z.

PLA-CF shrinkage

No data sheet checked for this page gives a printed-part figure for PLA-CF. Sunlu lists 0.3 to 1.1% without naming the method. The nearest printed-part data is plain PLA in the Binghamton study: 0.63% in X, 0.35% in Y and 0.80% in Z. Polymaker’s glass-filled HT-PLA-GF measures 0.05% in X-Y and in Z. Start from plain PLA’s values and measure.

ABS-GF and ABS-CF shrinkage

ABS-GF shrinks about 0.5%, per Sunlu’s product page and comparison table, against 0.4 to 0.8% for its plain ABS. Bambu Lab’s ABS-GF data sheet claims better dimensional stability than ABS but gives no percentage.

No ABS-CF data sheet checked for this page states a figure. Plain ABS in the Binghamton study shrank 0.59% in X and 0.69% in Y but grew 0.92% in Z. Measure the height of a filled ABS part for the same growth.

PETG-GF shrinkage

No PETG-GF data sheet checked states a figure. The nearest is Raise3D’s Industrial PET GF, a glass-filled PET: up to 0.1% Z shrinkage after annealing and no significant X-Y change. Use the PETG-CF range as a starting value and print a test part.

Why one XY shrinkage value is a compromise for fiber filament

Fibers line up with the bead. Tekinalp et al. at Oak Ridge printed carbon fiber ABS and found up to 91.5% of the fibers oriented in the printing direction. Along a bead they hold the plastic; across it, plastic does most of the shrinking.

In a long bar printed flat, the perimeters run along the bar, so its length is held by the fibers and its width is not. Turn the bar 90° and the low-shrinkage direction moves from X to Y. OrcaSlicer and PrusaSlicer have one XY field, and Bambu Studio has one shrinkage field covering X and Y, so any single value splits the difference.

When one dimension matters:

1. Calibrate flow first (dimensional accuracy calibration), so a fixed wall error is not read as shrinkage.
2. Print the part, or a 100 mm bar, in the orientation you will use.
3. Measure both directions and the height at room temperature.
4. Set the filament shrinkage to 100% (0% in PrusaSlicer) and scale X and Y separately in the object’s scale tool, with uniform scale off.

Say a 100 × 20 mm bar laid along X measures 99.70 mm long and 19.88 mm wide. That is 0.30% shrinkage in X and 0.60% in Y, so scale X to 100.30% and Y to 100.60%.

Holes need separate handling. The 10 mm hole in Polymaker’s PET-CF17 cube printed at 9.80 mm (2% undersized) and closed to 9.60 mm after annealing. That is far more than the cube’s 0.15% outer change, so correct holes with your slicer’s hole compensation, not the scale.