The day we bought a granulator and stopped pretending

The business case for this purchase was weak and I am going to show you the numbers, because the interesting part is why we signed it anyway.

The weak business case

A 275-gallon composite tote that fails resale grading yields roughly 52 kg of HDPE and 44 kg of galvanised steel. At the regrind and mill prices of the time, that material was worth about $31 at the gate. Granulating and baling it in-house cost us around $19 a unit in labour, power and amortised capital.

A twelve-dollar margin on six per cent of intake with a four-and-a-half-year payback is not a business. It is a rounding error with a maintenance schedule. Selling the units to a scrap contractor was simpler and nearly as profitable, and that is what we had been doing.

The board discussion, such as it was between three people in an office, came down to whether we were buying a profit centre or buying an answer to a question. We were buying an answer to a question.

What outsourcing actually bought us

Before 2017 mixed-condition units went to a regional scrap contractor on a weight ticket. They paid us, took them away, and that was the end of our involvement.

The problem surfaced when a customer asked a reasonable question: what happens to the ones you cannot resell? And the honest answer was "a contractor takes them, and I believe they are shredded and sold, but I cannot show you a document that proves it".

I could not show them because a mixed shred stream does not produce one. The contractor was not doing anything wrong — they were running a scrap yard, and scrap yards shred mixed material and sell it by weight. That is the job. But once the units left our gate, the chain of custody ended, and so did any claim we could make about them.

We were saying "nothing we touch is landfilled" and the true statement was "nothing we touch is landfilled by us".

That distinction sounds pedantic until somebody with an audit checklist asks you to evidence it. Then it is the whole difference between a disclosure and an assertion.

The difference the equipment makes

With a granulator and a baler on site, the disassembly happens under our roof and produces three separate, weighed, documented streams instead of one mixed one.

1. The bottle is cut out and granulated. The regrind is sorted by colour and melt index, which is worth doing because natural commands roughly 20% over mixed.
2. The cage is sheared and baled to the density and dimension the mill specifies. Baled galvanised steel is worth substantially more per ton than loose cages.
3. The base is triaged by material: reused if sound, repaired if economic, scrapped by stream if not.
4. Each stream crosses a certified scale and generates a ticket against the intake reference.

That last line is the whole purchase. The tickets are what turn a claim into a document, and the document is what we attach to a diversion certificate.

What we had to learn the hard way

Three operational lessons from the first year, none of which were in the equipment supplier's documentation.

• Strip everything before granulating. A single polypropylene valve body in a 500 kg batch of HDPE shows up as inclusions in extruded pipe and the batch gets rejected. We now remove valves, lids and gaskets as a separate station rather than a step.
• Residual product is a granulator problem, not just a safety one. Even a well-drained bottle holds a film. Granulating wet produces regrind that clumps in the hopper and a smell in the building that takes a week to clear. Units now get a rinse before they reach the cutting bench even though they are being destroyed.
• Bale density is a specification, not a preference. Our first loads went out under-compressed because nobody had asked the mill what they wanted. Loose bales are worth meaningfully less per ton and we gave away about $6,000 learning that.

The sorting discovery

One thing we did not anticipate: sorting regrind by colour and melt index turned out to pay for a meaningful share of the operating cost. Mixed-colour HDPE regrind goes into dark products — drainage pipe, agricultural ducting, pallet blocks. Natural regrind has more buyers and fetches more.

A scrap contractor has no reason to sort at that granularity, because their economics are about throughput. Ours are about the material, so we sort. The extra revenue narrowed the payback from 4.5 years to roughly 3.2, which nobody modelled in advance.

It also changed what we buy. Knowing that a batch of 200 identical bottles from one filler produces a consistent, saleable regrind grade means we now actively want homogeneous scrap intake, where previously we treated all scrap as equivalent. That shows up in the prices we quote for decommissioning jobs.

Where it sits in the hierarchy

I want to be careful not to oversell this. Recovery is the third-best thing you can do with a container, after reusing it as a container and reusing it as an object. Granulating a tote recovers the molecules and still pays the re-melting energy. It beats virgin material by a wide margin and loses to reuse by a wider one.

A granulator is an admission that the two better options failed. We run one because those options fail about six per cent of the time, and that six per cent still has to go somewhere we can point at. The full hierarchy, with the arithmetic, is on our eco mission page.