A dimension outside its expected range is a symptom, not a root cause. For an OEM, procurement or quality team, the immediate task is to determine whether the result represents a measurement issue, a stable but incorrect tool-to-part relationship, a cavity-specific condition, a time- or conditioning-related change, or process drift. This guide is intentionally focused on investigation and release evidence rather than a generic tolerance plan. Shrinkage and dimensional variation depend on the exact resin grade and lot controls, pigment or filler package, geometry, flow direction, gate and cooling design, mold condition, machine response, process window, post-molding history, and measurement method. ASTM D955 notes that standard-specimen shrinkage data cannot predict absolute values for production parts with differing flow paths, wall thicknesses, gradients and process conditions [2]. A useful buyer response links the observed symptom to comparable checks and a defined acceptance decision instead of treating every dimensional finding as a request to change pressure, temperature or steel.
Do not approve a correction because a single sample moved toward nominal. Confirm that the agreed measurement condition, affected scope, functional interface and any trade-off on other CTQs have been assessed.
01
1. Describe the symptom before assigning a cause
Start with a precise, neutral problem statement. State the part number and revision, feature identifier, nominal and limit, actual result, units, tool and cavity where known, sample quantity, date or lot, resin identification, and measurement condition. Add the direction and pattern: for example, every part is undersize at one bore; one cavity produces a larger center distance; a flatness result changes after 24 hours; or readings disagree between fixtures. ‘Excessive shrinkage’ is not yet a diagnosis. It may describe a size shift, while warpage, datum instability, gauge loading, or conditioning may be the mechanism that produced the reading.
Ask the supplier to segregate the finding into one of four practical patterns before proposing a remedy: a repeatable all-cavity offset, cavity-to-cavity variation, within-lot drift, or change after time, moisture exposure, temperature stabilization, or assembly. Compare the result with the approved baseline, not only the drawing nominal. A part can be within a broad drawing limit but still show a meaningful shift in a locating feature or mating condition. Conversely, a close-limit number should not trigger a tool change until the method and part state are confirmed.
The buyer question is: what acceptance requirement is actually threatened? It may be a direct dimension, a profile relationship, a mating gap, a seal, insertion force, cosmetic flushness, or a downstream process. Record that consequence. ISO 20457 provides a plastics-specific framework for tolerances and acceptance conditions because material behavior, shrinkage, processing, geometry, warpage and non-uniform cooling can affect part variation [1]. The standard does not turn a generic symptom label into a part-specific cause or acceptance decision.
02
2. Build likely causes across the full molding system
A repeatable size offset can be associated with resin behavior, actual packing history, mold dimensions, or the condition in which the part is measured. Resin family alone is insufficient: amorphous and semi-crystalline materials can respond differently, while reinforcement, orientation, colorants, regrind policy, moisture management and lot variation can change the part response. Covestro identifies material properties, filler content and amorphous or semi-crystalline behavior as relevant to shrinkage, alongside processing, wall thickness and part stiffness [3]. Require the exact approved grade and any permitted alternative to be identified before comparing results.
Geometry and mold effects can be local rather than global. Uneven walls, ribs, heavy intersections, long flow paths, asymmetric gates, restricted packing paths, changing cavity-surface temperature, vent condition, cooling balance, gate wear, ejector restraint and a damaged or modified insert can each affect an observed feature. Directional behavior is also important for filled materials and features aligned with, across, or near a changing flow path. A result that clusters by cavity or by a feature’s location should prompt a mold and flow-path review, not a blanket conclusion that the machine is unstable.
Machine and process contributors should be examined as a controlled set, not as a catalog of universal settings. A changed melt or mold temperature, transfer point, holding profile, effective gate-freeze behavior, cooling time, shot consistency, recovery, cushion, clamp behavior, material residence time, or drying condition can alter a molded part. The interaction depends on resin, geometry, mold and press. Ask which variables were recorded for the suspect lot, which stayed within the approved or trialled window, and whether the observed trend follows a documented process change.
03
3. Use confirmation checks that can separate competing explanations
Confirmation begins by checking the measurement system. Re-measure retained or fresh samples with the same controlled drawing revision, feature definition, datum alignment, fixture or support condition, instrument status, operator approach where relevant, and stated temperature or conditioning state. If the part is flexible or easily distorted, specify whether it is free-state, gently supported, or functionally constrained. The confirmation check is not merely a second number; it is evidence that the number answers the agreed requirement. A supplier should provide raw or traceable result records, sample and cavity identity, and an explanation of the method used.
Then use comparisons that isolate one question at a time. Compare cavities using samples from the same run and condition. Compare early and later measurements using the same method to assess post-molding change. Compare retained approved samples with current material and process records to distinguish a baseline shift from an isolated nonconformance. Where a feature’s function is the concern, add a representative mating, leak, fit, force or assembly check. NIST’s work on in-situ shrinkage monitoring illustrates that part shrinkage can be observed during molding; production acceptance still requires a method appropriate to the actual feature and requirement [4].
A credible cause statement should name both supporting and conflicting evidence. For example, a cavity-specific deviation supported by a cavity map, same-run material record, and repeatable measurement deserves a different response from a lot-wide shift that follows a verified material-drying or packing change. If evidence cannot separate two plausible causes, record the uncertainty and request the next targeted check. Do not represent a hypothesis as a confirmed root cause because it sounds familiar.
04
4. Correct the verified mechanism, then check for side effects
Corrective action should match the confirmed mechanism. A measurement-definition problem may require an agreed fixture, datum interpretation, conditioning instruction, or report revision rather than a physical part change. A verified material-control issue may call for restoring the approved grade, drying route, handling control, or approval path for a substitution. A process-related finding may justify a documented adjustment within the validated operating approach, followed by targeted samples. A cavity, gate, cooling or steel issue may need maintenance, repair or a controlled tool modification. None of these actions should be assumed suitable without the part-specific evidence.
Before changing the tool or process, require a stated correction hypothesis: which characteristic is expected to move, by what direction, what other CTQs could be affected, and what evidence will determine success. This is especially important when a correction intended to alter a local dimension could change flash, sink, residual stress, warpage, cycle time, appearance, or another cavity’s balance. The supplier should identify whether the proposal is reversible, whether it changes the approved baseline, and whether a new sample or functional review is needed before routine production resumes.
The confirmation of corrective action should repeat the agreed measurement condition and include the affected scope. A single favorable reading is weaker evidence than a documented comparison of relevant cavities, samples and functional checks. If a temporary deviation is commercially necessary, keep it separate from permanent approval: identify the quantity or lots covered, the acceptance authority, containment, expiry or review trigger, and the evidence still required.
05
5. Turn the investigation into a buyer acceptance record
For each dimensional incident, ask the supplier for an evidence package proportionate to the risk: the controlled drawing or CAD revision; feature and datum definition; resin grade and lot or batch identity where applicable; tool, cavity and tool-change context; relevant process-record excerpt; sample and measurement-condition identification; results by cavity or time point where relevant; suspected cause; checks performed; correction; and verification results. The package should make it possible for an overseas buyer to understand what was measured and why the supplier believes the action addresses the symptom.
Write acceptance criteria before reviewing the correction. Criteria may include all named CTQs meeting their limits under the stated measurement condition, no unexplained cavity-specific outlier, a completed functional assembly check, a documented visual or leak criterion, or a defined capability or monitoring study when the program requires it. The appropriate threshold, sample size and study design are product-specific. Avoid promising a universal Cpk, process setting, shrink allowance or inspection frequency where the product, risk and contractual requirements have not established one.
Finally, preserve comparability. Link the final decision to the material, tool state, process basis, measurement method and sample condition that were accepted. Define the change triggers that require review, such as material-grade or additive change, gate or cooling work, cavity repair, mold transfer, process-window revision, new measurement fixture, or a changed mating component. This record gives procurement a clear evidence request, engineering a traceable technical baseline, and quality a way to distinguish a contained exception from a released production condition.
Practical takeaway
Before you move
to the next review.
- State the symptom by feature, nominal and limit, direction of shift, sample/cavity identity, controlled revision, part condition and measurement method before asking for a root cause.
- Request evidence that separates all-cavity offset, cavity-specific variation, lot-to-lot or within-run drift, and post-molding or conditioning-related change.
- Confirm the exact resin grade, color or additive route, material-handling basis, tool and cavity context, and relevant process records for both approved and suspect samples.
- Require a correction hypothesis that identifies the targeted mechanism, anticipated effect, possible impact on other CTQs, verification method and approval decision needed.
- Release only against written acceptance criteria and retain the approved measurement condition, evidence, open actions and change triggers with the product record.
FAQ
Common questions
Can a material shrinkage value be used as the final correction for an injection mold?
No, not as a stand-alone final correction. Published material or standard-specimen values can inform early planning, but actual part behavior depends on the selected grade, geometry, flow, packing, cooling, mold and measurement condition. ASTM D955 specifically cautions that standard specimens do not predict absolute shrinkage values for parts with different flow paths, wall thicknesses, pressure and temperature gradients, and process conditions [2]. Confirm critical features on samples from the actual tool using the agreed method.
What supplier evidence should a buyer request when only one cavity is out of tolerance?
Request a cavity-identified dimensional map from comparable samples, the controlled measurement method and condition, tool and maintenance history, relevant process records, material identity, and the supplier’s inspection of cavity-specific contributors such as gate, venting, cooling, ejector restraint, insert condition or wear. The evidence should also show whether the pattern repeats after a targeted action and whether other CTQs moved.
When does post-molding measurement timing matter?
It matters whenever the polymer, part geometry or end-use condition can change the measured result after ejection. Define the time after molding, temperature and humidity or conditioning route, and whether the part is measured free-state or in a fixture. Compare results only when those conditions are aligned. The relevant timing is the one that supports the product’s acceptance or functional decision, not necessarily the most convenient time to inspect.
Should a supplier change process settings before a root cause is confirmed?
A controlled diagnostic trial may be appropriate, but it should not be presented as a proven permanent correction. Ask the supplier to state the suspected mechanism, the variable to be evaluated, the expected feature response, the samples and measurements to be taken, and potential effects on other CTQs. Where a temporary adjustment is used to protect supply, document the containment, scope, acceptance authority and follow-up evidence separately from final release.
References
Source material
- [1] International Organization for Standardization, ISO 20457:2026, Plastics moulded parts — Tolerances and acceptance conditions.
- [2] ASTM International, ASTM D955-21, Standard Test Method of Measuring Shrinkage from Mold Dimensions of Thermoplastics.
- [3] Covestro, The Fundamentals of Shrinkage in Thermoplastics.
- [4] National Institute of Standards and Technology, In-Situ Monitoring of Product Shrinkage During Injection Molding Using an Optical Sensor.
This guide is general information for early procurement and engineering discussion. It does not replace product-specific DFM, qualification, validation, regulatory assessment or legal advice.