100 mm is the length at which shrinkage stops being academic, so here is the same dimension across the materials people actually print:
| Material | Contraction | Finished dimension |
|---|---|---|
| PLA | 0.2–0.5% | 99.8–99.5 mm |
| PETG | 0.2–0.6% | 99.8–99.4 mm |
| ABS | 0.4–0.8% | 99.6–99.2 mm |
| PA6 | 0.8–2% | 99.2–98 mm |
The bottom row is ten times the top one. That is the whole reason material choice comes before tolerance design.
What each of those errors costs you
- PLA at 0.5% — half a millimetre. A hole pattern will not line up with a drilled plate, and a lid will be visibly tight.
- ABS at 0.8% — 0.8065 mm of correction needed. Enough that two mating parts printed in ABS still fit each other, while neither fits anything bought.
- PA6 at 2% — 2.0408 mm. Nothing designed to nominal will fit anything.
Note the second point, because it is the useful one: parts that only have to mate with each other tolerate large shrinkage, since both shrink together. Parts that have to meet a bought component do not.
Direction matters at this length
Across 100 mm the anisotropy is measurable. X and Y contract as the tables above describe; Z contracts less, because each layer is bonded to a stack already fixed to the plate.
Compensating uniformly at this scale produces a part that is correct on the bed and wrong in height by a visible margin.
The threshold in practice
Below about 40 mm, PLA and PETG shrinkage is inside the noise of a well-calibrated machine and can be ignored. Above 100 mm it cannot, for any material.
Between those, it depends entirely on whether the dimension has to mate with something. A decorative 60 mm part needs nothing; a 60 mm bracket bolting to a drilled chassis needs compensating.
The filled grades cut it dramatically
PETG-CF contracts 0.1 to 0.4% where plain PETG runs 0.2 to 0.6%, and PA6-CF contracts 0.2 to 0.6% where plain PA6 reaches 2%. Chopped fibre restrains the polymer matrix as it cools, and it narrows the band as well as lowering it.
For a large part that has to be dimensionally right, that is a stronger argument for a filled filament than stiffness ever is.
Compute the figure for your own dimension and material in the shrinkage calculator, and see per-material bands on the materials index.