Quality of the polymer, the batch and the decision: why the producer, the moulder and the end user rate different things

Most arguments about “good” and “bad” raw material are not arguments about facts. They are arguments in which the parties are rating different things. The word “quality” on its own proves nothing. Three things have to be named: what is being rated, against which requirements, and who set those requirements. In a polymer supply chain one word covers three separate objects — the individual batch, the grade itself, and the decision that joined the material to a design and a process. A batch can match its specification perfectly while the grade falls short for that particular part. A grade can have ample potential while a batch arrives with unacceptable spread. And batch and grade can both be sound — and the part still fails, if the selection decision was made with no margin. No incoming inspection repairs what was decided before the first bag of material arrived.

Three objects of assessment

It is more practical to start from what is actually picked up and rated.

The distinction settles several common disputes. Material produced ahead of demand, not yet contracted to a particular process, is rated against market medians or against the producer’s own portfolio. That is a measurement of technical level, not of conformity: there are no requirements yet, only a comparison with alternatives. A producer of recycled or virgin material who sincerely rates his grade as top is also speaking about technical level — within his own frame of comparison he is right. For a buyer with project requirements the same grade may fail the entry threshold, and that judgement is about a different object. Both are right, because they are rating different objects.

Who rates what along the chain

Prediction and responsibility part company in the first row. The compounder does predict how the material will behave in the moulder’s tool — from the formulation, the grade data and experience with similar parts. The part itself, however, comes out at the moulder, and responsibility for it stays there.

Translating requirements across the joints of the chain is joint engineering work. The customer is not obliged to name a polymer class or grade, but the reliability of the conditions depends on him: temperatures, loads, contact media, expected service life, consequences of failure. Turning those conditions into material properties, test methods and acceptance criteria is shared work: design owner, moulder and material supplier do it together. No party can translate conditions it was never told.

Three links is a simplification. In a real OEM chain the part customer, the design owner, the component manufacturer and the end user may be different parties, and responsibility for setting and verifying requirements is distributed among them. Material and product move along the chain in one direction, requirements move toward them, and at every joint a requirement needs translation.

Why there are several definitions of quality

The disagreement behind such disputes is older than the polymer industry. A supplier who writes “quality confirmed by certificate” is speaking Crosby; a moulder whose “batch is the same but behaves differently” is speaking Deming; a customer returning a product that “formally conforms but does not work” is speaking Juran. Each of the classic definitions still lives in somebody’s contract or complaint, and there is no winner among them.

David Garvin showed in his 1984 paper that these views formed independently of one another, in different disciplines — from philosophy to marketing and operations management. That is why they do not reduce to a common denominator.

What quality disputes actually come from

Four cases from market practice. In each: what happened, then what it was called, then what was actually being rated, and finally which of the three objects the question belongs to.

The housing cracks, the screws changed. The problem surfaced at assembly. The product was classed there as defective, and from there — automatically — as defective raw material: that is the shortest chain of inference, and it draws itself. Nothing in production had changed except the fastener batch: the screws arrived with a different diameter, and the change went unnoticed. The cause was found by checking the whole path of the product, not just the material. The cracking came not from a deviation in the polymer but from a change in the load imposed by the mating part.

The objection “a good material should withstand that” is partly fair. But the requirement to withstand variation in adjacent components is set at the selection stage, and a batch that never received that requirement cannot meet it. A selection decision carries not only the load from the product’s function, but also variation in the moulder’s other processes and components: tolerances of mating parts, fastener batches, changes in adjacent assemblies. The question belongs to the quality of the selection decision.

A batch was validated, not a specification. A moulder trials a new material and finds it suitable. The question is not whether he is right, but what exactly he validated. The batch used for the trial occupies a single point inside the specification range, and that point may sit close to an edge. What was validated is then the batch, not the specification: the next delivery, equally legitimate against the same specification, will sit at the opposite edge — and the margin that looked sufficient disappears.

This will be called a deterioration in batch quality. What was actually at issue is how representative the validation was. It is a property of the procedure, not anyone’s oversight: validation is run on the material physically available, and where that batch sits inside the range is usually not visible from the batch document — specification limits are often known only in part. The question belongs to the quality of the selection decision: the decision rests on a validation whose representativeness was never established.

There are two practical moves. Validating on two batches from opposite ends of the range shows whether the decision holds the whole range rather than one point. Agreeing a working range with the supplier instead of a single nominal moves the same understanding into the documents. This is done selectively: where the parameter genuinely matters for that product and where it can realistically be verified.

“Natural colour means better quality.” A common situation at the customer joint is a requirement phrased from experience rather than from calculation. Example: “natural colour” as a sign of quality, or “polyamide” with no critical conditions named. This is called a requirement on material quality. In reality there is a function behind it that colour performs in this project: it confirms the purity of the stream, makes contamination easier to spot visually, is needed for subsequent colouring, or matches a market expectation.

A colour requirement is a full part of the technical brief regardless of where it came from. Once the function is named, colour is either fixed as an acceptance characteristic in its own right, or separated from the properties the customer associates with it. The question does not belong to quality at all — it is a question of translating a requirement, and it is settled before the first batch exists.

The formulation was improved — and another product started failing. A compound formulation is changed with improved characteristics: a parameter that was a requirement of one product is raised specifically for it. The same material also sits on a second product, chosen by the moulder independently — and that product is sensitive to a change in the very same parameter.

This will be called a quality complaint against the material. In reality two different requirement systems were acting on one material: for the first product the change was an improvement against a direct requirement, for the second it was a change of the validated state. The compound became both better and unsuitable at once, and both ratings are honest. The question belongs to the selection decision for the second product: it was taken without the supplier, so there was no one to warn about the change. This is closed by two things — a more suitable grade for the second product, and an understanding that the moulder informs the supplier about new applications of the material.

All four are taken from the region where the object of assessment is hard to pin down. When a batch genuinely departs from its specification, the question is asked directly and settled by the batch document and a repeat measurement.

Quality of the selection decision

The third object is the most expensive one, because the decision is taken before the first batch and is rarely revisited.

The batch document lacks the essential thing — margin. A part does not fail from short-term loading alone: there is also creep, fatigue, low-temperature impact, weld lines, moisture uptake, ageing in the service environment. A selection decision holds margin against the mechanisms that are actually at work in that part. And two more things: tolerance for processing variation, and the match between grade, environment and service life.

When a decision is taken with no margin, the production run rests on a favourable coincidence of batch and cycle. The material that then looks “poor quality” is the one behaving faultlessly on other projects. A batch should not accidentally compensate for insufficient margin in the decision: a run has to work across the whole agreed range of variation.

Telling “the batch is drifting” from “the decision has no margin” is helped by diagnostics across three linked data sets: material indicators, process response, part characteristics. Correlation between them does not yet prove causation. The cause is confirmed by a controlled comparative run under unchanged conditions, a verified measurement system, and clear batch identification in the records.

What to agree before a production run

1. Function, environment and service life of the part — temperatures, loads and their cycling, contact substances, moisture, expected life, consequences of failure.
2. Critical characteristics of the product — geometry, mechanics, tightness, electrical parameters, appearance: what actually determines fitness.
3. Decision margin and processing window — against the relevant failure mechanisms, with allowance for processing variation; limits of moisture, melt and tool temperatures, pressures and times within which the process is approved.
4. Acceptance characteristics and validation plan — only those indicators for which a link to the process or the part function is justified, with method and sampling point; separately, what is checked on a specimen, what on the real part, and in which conditioned state.

And one more understanding that belongs to none of the points above: which language deviations will be discussed in, once they appear. That is, which of the three objects is on the table — the batch, the level of the grade, or the selection decision.

Expert review: 5 questions

1. Why not adopt a single definition of quality and use it? Each of the classic definitions answers its own question: conformance is about doing what was written, fitness for use is about working in real conditions, uniformity is about stability, the loss function is about the cost of deviation. The first governs a supply contract, the second a user complaint, the third a series process. Choosing one means losing the remaining questions, which do not disappear because they stopped being asked.

2. A supplier calls a material high quality — what stands behind that? Most often an honest comparison with his own portfolio or with market medians. A producer whose new grade beats all of his previous ones is right within his frame of comparison. The issue is not sincerity but the object: compared against what, on which indicators, by which methods. Once the object is named, the rating becomes a testable statement about technical level, and it can be set against the requirements of a specific project.

3. A customer asks for “natural colour, because it is better quality”. Argue? A colour requirement is a full part of the technical brief regardless of its origin. The productive route is to establish what function colour serves in this requirement: control of stream purity, ease of spotting contamination, preparation for colouring, or a market expectation. Colour then either becomes an acceptance characteristic in its own right, or is separated from the properties the customer associates with it — and those are confirmed by their own methods.

4. Batches conform to specification, but the part is unstable. Who is responsible? At the diagnostic stage a question about mechanism works better than a question about blame: three data sets — material, process, part — show what the instability correlates with. If the part drifts together with batch drift, the discussion is about acceptance limits with the supplier; if there is no drift, it is about the process or the decision margin. Once the root cause is confirmed, the second part begins: contractual responsibility, corrective actions, disposition of product.

5. Can a material be high quality “in general”, with no project attached? Without requirements only technical level exists: indicators against market medians or against an own portfolio. That is a correct and useful assessment — it is what a new product launch relies on — but it answers the question “how good is this material among comparable ones”. The question “will it give a stable part in your process” has no meaning without a project.

Summary

The word “quality” covers three different objects: the batch, the technical level of the grade, and the selection decision. The first is proven by documents and measurements, the second by comparison with alternatives, the third by part validation and performance margin. Disputes about “quality” usually turn out to be unreconciled objects of assessment, and the shortest way out is to name which of the three is under discussion and against which requirements. The costliest confusion in the market is between the first and the third: a batch should not have to compensate for a decision taken with no margin.

Examid® PA6 GF30 — Glass-filled polyamide 6 with 30 % glass fibre, base grade for series injection mouldingSpecification and batch document · agreeing acceptance indicators for the projectRequest supply terms →

Examid® engineering polyamides — Portfolio hub: unfilled, glass-filled and carbon-filled gradesGrade selection against part function, environment and processing windowGo to the catalogue →

Material Wizard works on both translation joints: turning application conditions into material requirements, comparing candidate grades by technical level, defining the critical acceptance indicators, and building a programme to verify the material in the real part and the real process. The final decision is taken together with the design owner and the moulder — on the basis of validation, not TDS data alone. To buy with delivery across Ukraine, or to check your project requirements against grade characteristics, ask a specialist.