A customer sent us a 3D file from a mold that had already run. The only request was to swap PC/ABS for ABS. We released the model as it was. At T0, all four corners of the housing had cracked, and the snap fits broke on the first latch.
Wall thickness was not the problem. Ribs were not the problem. Every dimension in that file had been scaled for the old resin's shrinkage. Changing the resin without changing the shrinkage is walking a new road with an old map. The mold was cut wrong, the parts came out wrong, and the whole T0 cycle had to be paid for twice.
The fix is one action, applied everywhere: rescale every dimension on the 3D model to the new resin's shrink rate, and let the model carry that value out to the mold shop. It is not a line in the BOM.
Shrinkage is not a number in the corner of the drawing
Most structural engineers open a drawing, find the technical requirements box in the lower right, and change "shrinkage 1.005" to "1.006". That is not wrong. It is also not enough. The shrink rate is the basis for every cavity dimension in the tool. The cavity is cut oversized by that factor. Change the factor and the cavity changes. If the 3D model was never rescaled, the mold shop machines to the old geometry and the parts come out at the wrong size no matter what the box says.
The numbers look harmless until you put a length on them. PC/ABS runs 0.4% to 0.6%. ABS runs 0.4% to 0.7%. On paper the gap is 0.1%. On a 200 mm housing, 0.1% is 0.2 mm. That is enough to break assembly at snap clearance, at boss center distance, and at the shutoff fit.
| Material | Shrinkage | What it does to the drawing |
|---|---|---|
| PC/ABS | 0.4% to 0.6% | Baseline for the original file |
| ABS | 0.4% to 0.7% | Wider range, batch spread can reach 0.1% |
| ABS + 10% glass fiber | 0.2% to 0.4% | Strong anisotropy between flow and cross-flow |
| PA, POM | 1.5% to 2.5% | PA keeps shrinking after ejection, 0.1% to 0.3% |
There is a second layer. Different resins distribute shrinkage differently through the wall and along length versus width. ABS is more anisotropic than PC/ABS, so the stress concentration points at wall thickness steps move as well. A drawing scaled uniformly for the old resin puts those points in the wrong place.
We had a project where the customer moved from PC/ABS to ABS with 10% glass fiber. Glass-filled shrinkage is 0.2% to 0.4%, roughly half of plain ABS. The drawing was never revised. The housing came out 0.5 mm short, the snap positions no longer lined up with the front cover, and the entire mid-frame was scrap. That 0.5 mm was not a tolerance issue. Every dimension was small in the same direction. Systematic offset, not random variation.
The drawing shows the finished part, not the mold
A 3D file describes the shape and size of the molded part. It does not describe the cavity. The mold shop takes the file, multiplies each dimension by the shrink rate, and cuts the steel. Give them the wrong rate and they apply the wrong multiplier. The logic is simple, yet most material changes stop at renaming the resin in the BOM.
The correct move is to rescale the whole part inside the 3D software. If the original file carries 1.005 and the new resin needs 1.007, scale the model by 1.007 / 1.005, about 1.002. This is not editing a text field. It means resetting the scale factor inside the modeling feature tree so the exported model itself carries the right value. Different CAD packages put the command in different menus. The principle does not change.
We got this wrong once. We updated the technical requirements box and left the model alone. The mold shop built the cavity to the box, but their programmer read dimensions straight off the 3D model and never opened the note. The cavity was cut at the old shrinkage. The part came out 0.3 mm small, and the snap clearance opened from 0.1 mm to 0.4 mm. The latch would not hold.
Three questions before you change material
When a material change request lands, do not start editing drawings. Ask three questions first.
What is the actual shrinkage? Get measured data from the supplier, not the range on the datasheet. ABS is listed at 0.4% to 0.7%, and batch-to-batch variation alone can reach 0.1%. The same grade from a different lot shrinks differently. Ask for the measured value for the lot, or shoot a sample from the pellets and measure it.
How anisotropic is it? Glass-filled grades shrink far less in the flow direction than across it. On a rectangular housing with the long edge along flow, the long and short edges need different compensation. A single global scale factor only covers one direction. Anisotropy has to be compensated by the mold shop during cavity machining, and that instruction has to be on the drawing.
How much post-shrinkage? Some resins keep shrinking after ejection. PA can move another 0.1% to 0.3% and only stabilizes after 24 hours. On a precision fit, that movement opens the gap. Leave allowance on the drawing or build the correction into the tool.
Shutoffs and snap fits are the most sensitive features
The places where shrinkage difference bites hardest are the shutoff and the snap fit. Shutoff clearance is typically 0.05 mm to 0.15 mm. A 0.1% shrink difference across a 100 mm shutoff length is 0.1 mm, which can consume the whole clearance. A straight shutoff can still be saved by reworking the tool. A tapered shutoff gives you very little rework range.
Snap fits are worse. The deflection available on a snap is usually only 0.3 mm to 0.5 mm. A 0.2% shrink difference across a 50 mm snap arm is 0.1 mm. The arm has almost no margin to absorb that. We have seen a material change make a snap too tight, and the latch snapped off during assembly. Teardown showed the arm had come out 0.15 mm shorter than expected, which pushed engagement depth up by 0.15 mm and drove snap stress past its limit.
So after any material change, check shutoff and snap dimensions separately. If the shrink rate moves by more than 0.1%, recalculate the fit clearance and adjust arm length or shutoff depth if needed.
A material change is not a renaming exercise. Every dimension in the drawing was designed around one specific shrink rate. Change the resin and the model has to be scaled again. Do the work up front. Fixing cracks after T0 costs double in tooling and in schedule.
FAQ
Where do shrinkage differences actually show up?
Three places: the shrink value itself, anisotropy, and post-shrinkage. Crystalline resins such as PA and POM shrink heavily, around 1.5% to 2.5%. Amorphous resins such as ABS and PC shrink little, around 0.4% to 0.7%. Glass-filled grades add strong anisotropy, with flow and cross-flow shrinkage sometimes differing by a factor of two. Matching the headline number is not enough. Match the material class too. Moving from ABS to PC/ABS keeps the number close, but PC post-shrinks more than ABS. On precision fits, mark the drawing "scale to measured shrinkage" and attach the supplier's measured data.
I changed the shrinkage in the 3D file. Can the mold shop still get it wrong?
Yes, if only the technical requirements box was edited and the model itself was not rescaled. Programmers usually read dimensions directly from the 3D model. If the model still carries the old shrink rate, they program the old size. A written note can be missed. The reliable route is to reset the scale factor in the CAD features so the exported model carries the correct value, then print the shrink rate in a visible spot next to the dimension and tolerance table with the note "this shrinkage is already applied in the 3D model". Confirm the number with the mold shop in writing when you release the files.
Which features fail first after a material change?
Shutoff fits, snap fits, boss center distance, and the corners of long housings. Shutoff and snap clearances are typically 0.05 mm to 0.15 mm, so a 0.1% shrink difference can erase the gap or double it. Corners on long housings concentrate stress, and a shrinkage mismatch raises local internal stress until cracks appear. Boss center distance affects screw assembly: two bosses 100 mm apart shift 0.2 mm at a 0.2% shrink difference, and a screw will skew in a blind hole. Run a separate dimension chain calculation on long dimensions, mating faces, and stress concentration points to confirm every clearance still sits inside tolerance.
How do I get accurate shrinkage data?
Ask the supplier for the measured shrinkage of the specific lot, or mold a sample from that lot and measure it yourself. Datasheet values are ranges, and real batch variation reaches 0.1%. For tight-tolerance parts, measured data beats the datasheet every time. To measure, injection mold a simple bar 200 mm long, 20 mm wide, 2 mm thick, measure its length, and calculate shrinkage as (cavity length minus bar length) divided by cavity length, times 100%. Measure 24 hours after ejection so post-shrinkage has fully released.
Sources: Compiled from Hezi Industrial Design's own project experience; no external citations.
Article by Hezi Industrial Design — Dongguan, China. Originally published at hezidesign.com.
Top comments (0)