Sink Mark Prediction and Prevention — How Deep Will a Sink Mark Be, What Depth Is Acceptable on a Cosmetic Surface, and How Do You Specify and Measure It Before the Mold Is Cut?
Last updated: 2026-09-03 · Reading time: about 6 min
Sink is a dimensional deviation you can put a number on before any steel moves. Estimate the depth from the local shrinkage differential, write a maximum of about 0.03 mm on a Class A surface into the drawing, and verify it on the molded part. A quantified prediction says “roughly 0.1 mm here — inside or outside your limit” instead of “this rib will sink.”
See also related article.
Key facts
<>
Warpage — Part deformation after ejection caused by uneven shrinkage. Amorphous resins warp less (0.4–0.7% shrinkage). Semi-crystalline resins warp more (1.0–2.5% shrinkage). Fix: uniform wall thickness, balanced cooling, optimized gate location, adequate packing pressure. See also related guide.
How deep will it be — turning a location call into a number before you cut steel
A qualitative review tells you a rib will sink — enough to start an argument, not enough to defend a spec. You can put a number on it before steel moves, with two tools that should agree: a hand screen from the shrinkage differential, and the simulation’s volumetric-shrinkage plot. Neither replaces a T1 measurement; both replace the opinion.
| Feature and driver | Amorphous (ABS, PC, PS) | Semi-crystalline (PP, PE, POM) | Glass-filled (~30% GF) |
|---|---|---|---|
| Rib, base ≈50% of wall | 0.02–0.05 mm | 0.05–0.12 mm | 0.01–0.03 mm |
| Rib, base ≈75% of wall | 0.05–0.12 mm | 0.12–0.30 mm | 0.03–0.08 mm |
| Solid boss, base ≈2× wall | 0.08–0.20 mm | 0.20–0.45 mm | 0.05–0.12 mm |
| Thick-to-thin step, 2:1 | 0.03–0.08 mm | 0.08–0.18 mm | 0.02–0.05 mm |
Read this as an order-of-magnitude screen, not a tolerance. The ratios describe geometry you already have — what ratio to choose is a separate design decision; this only tells you the depth a given ratio produces.
The math is arithmetic: depth scales with the differential volumetric shrinkage between the thick feature and the nominal wall, times the span of surface it pulls in. Semi-crystalline resins run roughly 2–4 times deeper than amorphous on the same geometry, and glass fill cuts that 60–80 percent — enough to separate a 0.05 mm zone from a 0.5 mm zone, not to carry the third decimal.
- Run volumetric shrinkage, not just fill. A fill study says whether it is short; the shrinkage result colors each node by how much material is missing at freeze-off. The hot spots are your sink locations, with a magnitude attached.
- Read the peak at the flagged node. Take the maximum volumetric shrinkage — or the sink index — and scale it by the local wall for a first-order depth. If it disagrees with the hand screen by an order of magnitude, one is wrong.
- Threshold it against the surface class. Compare the predicted depth to the drawing limit for that zone. Under it, the zone passes on paper; over it, flag it for a design or process change before the tool is cut, not after T1.
- Correlate predicted to measured at T1. Measure the real depth on the first shots and plot it against the prediction. The gap — packing profile, gate freeze, resin-lot variance — sets the margin you carry on the next zone.
How deep is too deep — specifying an acceptable sink depth and measuring it
A prediction is only useful against a limit, and the limit belongs on the drawing as a maximum sink depth per surface class. It also has to be verifiable with an instrument and a stated reference — otherwise “acceptable” is an argument, not a spec.
| Surface class | Where it sits | Typical max sink depth (starting spec) | Verify with |
|---|---|---|---|
| Class A — high-gloss cosmetic | Visible A-surface, polished or gloss | ≤0.03 mm (0.0012 in) | Profilometer or 3D scan |
| Class B — textured or matte cosmetic | Visible B-surface with a texture | ≤0.08 mm (0.003 in) | Dial indicator or profilometer |
| Class C — non-cosmetic or hidden | Internal, back side, paint-covered | ≤0.15 mm (0.006 in) | Height gauge or boundary sample |
These are starting points, not law — adjust every value to how the part is seen and used, and say so on the drawing. A limit with no measurement behind it is decoration.
Sink depth. The perpendicular distance from the intended, un-sunk surface to the lowest point of the depression, measured on a part at room temperature after shrinkage is complete — not at ejection, when it is still hot and moving. Fixing that definition, and the reference it is measured from, is what makes a 0.03 mm spec repeatable between two people and two instruments.
“‘This rib will sink’ starts an argument. ‘We predict about 0.1 mm here, your Class A limit is 0.03 mm, so we propose coring the boss — and here is the T1 profilometer trace that confirms it’ ends one. A prediction earns trust only when it is a number, checked against a written spec, then measured on the part.”
— Project Engineer, aCatMould
Frequently asked questions
How close is the predicted sink depth to what I measure on the part?
Close enough to decide, not close enough to skip the measurement. The hand screen and the simulation usually land within the same order of magnitude as the molded part — they reliably separate a 0.03 mm zone from a 0.3 mm zone, and that is the whole call.
The residual gap comes from packing profile, gate freeze, cooling balance around the thick zone, and lot-to-lot shrinkage variance. That is why you correlate prediction to measurement at T1 and carry the gap as margin.
How do I measure sink depth on a textured surface with no flat reference?
You lose the straight-edge datum, so you change the reference. Best-fit a 3D scan of the textured surface to the CAD and read the deviation from the intended surface — the texture averages out over the scan, the sink does not.
Or measure against an un-sunk area of the same texture, or a physical sink boundary sample approved for that zone. State which reference the limit uses, or two people will measure the same part two ways and both be right.
Who owns the acceptable sink depth — me or the mold maker?
The limit is yours — it comes from your product’s appearance and function. You set the surface class and the maximum depth on the drawing; the mold maker predicts against it, proposes the design and process to hit it, and verifies with a measured report at T1.
If your program needs a documented sink-depth verification, tell us before the deposit and we will say plainly which parts we can support with measured data and which sit with you.
Key takeaways
- Quantify sink before steel moves: a shrinkage-differential hand screen refined by the volumetric-shrinkage plot gives a predicted depth and location, verified at T1.
- Give the prediction a limit — a maximum sink depth per surface class on the drawing, roughly ≤0.03 mm Class A gloss, ≤0.08 mm textured, ≤0.15 mm non-cosmetic.
- Specify the depth against a stated reference and measure it — height gauge, dial indicator, profilometer or 3D scan — on a room-temperature part after shrinkage is complete.
Sink prediction is not an opinion about where a dent might show up — it is a number, a written limit, and a measurement that closes the loop. Estimate the depth, refine it with the simulation, specify a maximum per surface class, and verify it on the part. At aCatMould we run that estimate against your limit before the deposit and hand you the measured T1 report.
Sources
Autodesk Moldflow — volumetric shrinkage and sink prediction
MatWeb — material property database (molded shrinkage by grade)
DuPont — injection molding part design guidelines
Plastics Technology — sink marks and dimensional molding
Society of Plastics Engineers — part design and measurement guidance
Sources: Plastics Technology · Plastics Technology Online · Society of Plastics Engineers (SPE) · ThomasNet