the tonne you paid for twice: why off-spec is a margin problem before it is a quality problem

Most compounders can tell you their scrap rate. Fewer can tell you what off-spec material costs them, because the cost does not sit in one place. Some of it is sold on at a discount, some is reworked into later batches, some waits in the warehouse for a customer who will never order that grade again, and some arrives back as a complaint. Each piece looks tolerable on its own; together they are a persistent drag on margins that are already thin.

The uncomfortable part is that a good share of off-spec is not bad luck. It is process variation that was measurable while it was happening, and simply was not measured, or not acted on, until the batch was finished.

where the money goes

Off-spec compound has already consumed everything good compound consumes: polymer, additives, energy, line time and labour. The loss is not the tonne; it is the gap between what it cost to make and what it can now earn. That gap shows up in a few familiar ways.

• Selling it on for less. The obvious route is to sell the material as a lower grade or as scrap. As one engineering consultant put it in Plastics Technology, reworking instead means the compounder "doesn't have to sell off the off-spec material as scrap at a lower cost" (Plastics Technology).
• Reworking it slowly. Rework is the better option, but it is not free and it is not fast. The same article suggests starting unproven blends at a 1–3% rework rate and building up gradually, and cites a 5% rate as an example from experience (Plastics Technology). At those rates, a large off-spec lot occupies warehouse space for a long time.
• Processing scrap at all. As another Plastics Technology contributor observed of regrind, "processing the polymer into scrap costs even more than processing saleable product" (Plastics Technology).
• Customer exposure. Off-spec that is caught is a cost. Off-spec that is not caught is a claim, a returned delivery or a customer quietly qualifying a second supplier. We have not found reliable public figures for the cost of claims in compounding, so we will not invent one; most commercial directors have their own number, and it is rarely small.

The compounder's position is captured neatly in that first article: "you are selling a specification, not a particular grade" (Plastics Technology).

recyclate makes the problem harder

For compounders and recyclers working with post-consumer material, the input is itself a source of variation. A 2022 paper in the journal Polymers notes that "post-consumer recyclates often exhibit strongly fluctuating MFR values from one lot to another", and that suppliers therefore "often only provide a wide range for the MFR in their data sheets" (Traxler et al., Polymers, 2022). A recent AMI feature describes the same challenge for companies reprocessing post-consumer material "with all its variability in quality" (AMI).

Converters notice. In a survey by European Plastics Converters, almost 60% of converting companies said they found it hard to obtain recycled material of acceptable quality (EuPC). That survey dates from 2017, so treat it as a long-standing complaint rather than a current measurement, but we see little evidence that the complaint has gone away.

And the commercial backdrop is not forgiving. Plastics Recyclers Europe expects almost one million tonnes of European recycling capacity to have been lost between 2023 and the end of 2025, including Biffa's plant in Sunderland (S&P Global). A recycler with consistent output has an argument for price; one without has an argument with its customers.

the lag between drift and discovery

Much off-spec is created in the gap between a process changing and someone finding out. A melt flow or property test in the lab is, by nature, after the fact. One rheometer supplier, describing the case for in-line measurement, puts the delay at "30 minutes to hours", adding that "by then, you may have made scrap for the entire intervening period" (Plastics Technology). That is a vendor making its case, but the arithmetic holds: at any meaningful throughput, an hour of undetected drift is a lot of material.

The causes are rarely mysterious. The same source lists "regrind/recycle content, additive package drift or degradation from thermal history" as things that shift viscosity (Plastics Technology). On regrind specifically, "the higher the percentage of regrind usage and/or its variability, the higher the scrap rate", and when the regrind share keeps changing, "operators are continuously required to change temperatures" (Plastics Technology).

In other words, the line is usually telling you something. Melt pressure, melt temperature, motor load and feed rates move before the lab result does. The problem is that on many lines those signals are visible only on a local panel, interpreted by whoever happens to be standing near it.

Academic work has explored this for some time. A 2017 SPE ANTEC paper, for example, used inline process data from a twin-screw extruder to predict product properties and flag out-of-specification material, with the stated aim of reducing scrap (Mulrennan et al., SPE ANTEC 2017). The idea is not new; making it work reliably on a production floor is the harder part.

what earlier detection is worth

We would be wary of anyone promising to eliminate off-spec. The more modest and more useful aim is to shorten the time between a process starting to drift and someone doing something about it. That is what we are working on at Krayo:

• Measuring process behaviour in real time, from the data the line already produces, rather than waiting for the lab to confirm what the extruder knew an hour earlier.
• Flagging drift early, while it is still a correction rather than a downgrade.
• Recommending setpoint adjustments to the operator, who stays in charge and decides whether to act.
• Letting a skilled operator supervise more lines, which matters when experienced people are the scarcest resource on site.
• Working with older machines that were never built with data connections in mind, because that describes a large share of the equipment actually running in UK compounding.

None of this is proven at scale yet, and we will say so plainly until it is. But the economics are simple enough to check on your own numbers: take last year's downgraded and reworked tonnage, multiply by the gap between prime and achieved value, and ask how much of it began as drift that was visible on the line before it was visible in the lab.

Krayo is building process intelligence for plastic compounders and recyclers. If that sum produces an uncomfortable number, we would be glad to hear from you.