Often when you ask a manufacturing site how much waste it produced last month and you will get an answer. Ask what it cost to remove and you will get that too, often to the penny, because somebody had to approve the invoice. Ask where the material came from, why it stopped being useful, and whether anything more sensible could have been done with it, and the conversation slows down.
That pause is not a failure of effort. Most engineers I meet care a great deal about the material leaving their site. The difficulty is that our waste records were designed to answer financial and compliance questions, not engineering ones. We know the tonnage. We rarely know the story behind it.
What our waste records actually record
Waste documentation in most operations serves two purposes: showing that duty of care has been discharged, and controlling cost. A transfer note tells you who collected the material, what broad category it was assigned to and where it went. The finance system tells you what that cost.
The explanation usually exists somewhere. It sits in a scrap code in the Manufacturing Execution System (MES), a non-conformance report, a maintenance entry about a worn tool or a drifting temperature, a supervisor's spreadsheet, a photograph on somebody's phone. Each fragment is owned by a different function, held in a different format and kept for a different length of time. By the time the skip is collected, the link between production event and disposal event has been lost.
The result is an odd asymmetry. We measure waste with reasonable precision as a mass and a cost. That makes prevention harder than it should be, and it leaves downstream recovery guessing. A recycler receiving polymer offcuts has little idea whether the material was contaminated by a release agent, reprocessed twice already, or rejected on a dimensional tolerance and otherwise sound.
A passport for the waste event
One way to close that gap is to give a significant waste stream or waste event its own digital record: a Waste Passport. Rather than a static certificate, it would accumulate information as the material moves, and might carry:
- material type, composition and physical condition;
- quantity and an estimate of residual value;
- the production stage and location where it arose;
- the reason the material became waste;
- relevant process, quality or maintenance conditions at the time;
- contamination status and handling requirements;
- candidate routes: reuse, repair, remanufacturing, recycling, recovery or energy recovery;
- the economic and environmental implications of each route;
- the final destination and what actually happened to it.
None of these elements are unique or exotic, and most of these functions or data exists somewhere in a modern factory already. What is missing is the discipline of assembling them at the moment the material is written off, while the context is still recoverable, and carrying them to whoever handles it next.
Not a rival to the Digital Product Passport (DPP)
It would be easy to read this as a competitor to the Digital Product Passport. It is not, and the distinction matters.
A Digital Product Passport describes the product. A Waste Passport explains the waste event.
The Digital Product Passport follows the product through its lifecycle. The Waste Passport begins when a material is rejected, discarded or diverted from its intended use. The two intersect at end of life, but a great deal of industrial waste never becomes a product at all. Offcuts, purge material, expired stock, failed batches, filter media and rework that could not be saved arise inside the factory and never reach a customer. They sit outside the product passport frame entirely, and are often the streams with the greatest recovery potential, precisely because they are relatively homogeneous and their history is known at the point of generation.
A Waste Passport is therefore a complementary data layer connecting manufacturing operations to downstream circular and resource-recovery systems, not a duplicate of what a product passport already does.
Where AI could earn its place?
This is worth discussing now because the linking problem has become tractable. Connecting a waste event to the conditions that produced it means reconciling timestamps, batch identifiers, asset numbers, shift patterns and a good deal of inconsistent free text, which is the kind of untidy work machine learning handles better than manual analysis.
Three uses look promising. The first is normalisation: turning inconsistent scrap reasons, local shorthand and operator notes into a classification that can be analysed across lines and sites. The second is pattern detection: identifying that a defect cluster tends to follow a particular changeover, tool life threshold or material lot, which moves the discussion from tonnage reporting towards probable cause. The third is route suggestion: matching the recorded condition of a stream against plausible recovery options, and flagging where material is heading for disposal when its properties suggest something better is available.
The point is not better documentation of waste after the fact. It is prevention first, and better decisions about the waste that cannot yet be designed out.
Where factory intelligence meets recovery
This is the question behind a conversation between my own organisation (Inshira) and Circular Bioenergy (CirBE). Inshira Technologies works on manufacturing intelligence: understanding where operational waste originates, connecting it to production decisions, identifying recurring causes and supporting prevention. CirBE works on the other side of that boundary, on circular resource recovery and the selection of appropriate downstream pathways including reuse, recycling and, where suitable, energy recovery.
Putting the two perspectives side by side produces a straightforward engineering question: can information generated inside the factory remain useful after the material leaves the production process? We are exploring the concept and what a practical implementation might involve. We have not built a Waste Passport platform, and it would be premature to suggest otherwise. The Founder of CirBE, Van Nguyen is quoted saying:
"Waste is often treated as the end of a process, when it should be treated as the beginning of another value chain. Better information about its origin, composition and condition could help organisations identify more viable routes for reuse, recovery and energy generation."
The awkward questions
An idea like this deserves scrutiny before enthusiasm, and there are several ways it could fail. Data quality is the obvious one. A passport assembled from inconsistent production records will be confidently wrong, which is worse than being visibly empty. Many sites would need to improve basic scrap capture before any of this became meaningful.
Then there is classification. Existing waste coding schemes were written for regulatory control, not recovery decisions, and there is no agreed vocabulary for describing condition, contamination or residual value in a way a downstream processor could act on. Without interoperability, every implementation becomes another silo.
Ownership is unresolved. Does the passport belong to the producer, the carrier, the recovery operator or a neutral registry? Who maintains it once the material has changed hands twice, and who is liable if it is wrong?
Commercial sensitivity is a real constraint, not an excuse. Waste data reveals yield, process stability and sometimes formulation, so any workable scheme needs graduated disclosure, where a recycler sees material condition without seeing the production performance behind it.
Verification matters more than it tends to get credit for. An AI-generated recommendation about a recovery route is an inference, not a fact, and composition claims travelling with a material may carry legal and safety weight. Something has to sit between the model's suggestion and the decision taken, whether that is testing, professional judgement or both.
There is also a cost test many streams will fail. If characterising a load costs more than the recovered material is worth, the passport is administrative burden dressed as sustainability. The idea has to earn its place stream by stream.
Finally, and to my mind most importantly, there is a moral hazard. Making recovery easier can quietly make waste more acceptable, and a well-documented stream with an efficient onward route starts to look like a success rather than a symptom. The waste hierarchy exists for good reason. Energy recovery has a legitimate role for materials that cannot practically be reused or recycled, but it should not become the default simply because it is the easiest option to arrange. If a Waste Passport does not make prevention more likely, it is not doing its job.
A set of open questions
I am not convinced the answer is a new system. It may be a shared data layer, a set of conventions, or a better handshake between systems we already run. What I am confident about is the gap itself. There is a point where a material stops being work in progress and becomes waste, and at that moment we lose almost everything we knew about it. Rebuilding that knowledge afterwards, by inspection and inference, is slower and less accurate than carrying it forward.
So I would put this as questions rather than proposals.
- Do engineers need a better information bridge between the point where material becomes waste and the point where it becomes a resource again?
- What minimum information would make a Waste Passport genuinely useful, as opposed to merely comprehensive?
- Who should be responsible for creating, verifying and maintaining it, and who carries the liability when it is wrong?
- Could it complement existing product-passport and waste-tracking obligations without adding a layer of administration that costs money and changes nothing?