In a specification review, the choice between two joining methods is normally settled on assembly hours, sealing performance and unit cost; a bonded or laminated joint, requiring fewer labour hours, fewer penetrations and a narrower warranty argument than a mechanically fastened alternative, is a defensible selection, and no one seated at that table holds a brief against it. Nowhere in the same meeting is there a budget line carrying the question of how that joint will be separated at the end of the asset's life, because decommissioning belongs to a cost centre that does not yet exist and will, in most cases, be managed by someone who is not employed by the same company. The decision is taken from inside a two-to-three-year commissioning schedule; its consequence appears on the books of the third or fourth owner, twenty to thirty years later.
The same pattern is more sharply visible on the procurement side. When a component bid is evaluated, the material declaration — alloy grade, coating chemistry, joining method — is generally treated as a technical annex rather than as a contractual deliverable, and even where the information sits in the tender file, carrying it forward into the as-built set requires a discipline that is, on most projects, the first link to slacken during the closeout push. At end of life the site prices not the design intent that was once described to it but only what can be observed: a decommissioning contractor builds a bid from the difference between the scrap value of the streams it can separate and the labour it will spend separating them, and widens that difference in its own favour wherever it encounters uncertainty. An undocumented alloy is, in practice, the lowest alloy in its family.
The shared name for these two observations is the circularity gap — the distance between a material being recyclable in principle and being economically recoverable inside the configuration into which it was assembled. What produces the distance is not chemistry but separation cost; recovery happens at the point where the market value of the reclaimed material exceeds separation labour, transport and requalification, and below that point the material remains technically recyclable while becoming, in operational fact, waste. A glass-fibre-reinforced thermoset blade, an encapsulated module laminate, a cell bonded into a pack housing, painted and galvanised steel entering a mixed stream alongside aluminium — the arithmetic is identical in each case: the material is present, the separation economics are not.
Recognising that this choice is not an error matters, because the problem has to be picked up from the correct end. Irreversible joints reduce mass, lower sealing risk, compress assembly labour and improve fatigue behaviour, which is to say they directly support the performance undertaking written into the contract and the warranty exposure sitting behind it. The decision is rational within the horizon in which it is assessed, and it remains rational for as long as one stays inside that horizon. The difficulty lies neither in the shortcut nor in the people taking it, but in an evaluation window sized at roughly a tenth of the asset's life, with no role at any stage of the project carrying the asymmetry.
A second mechanism widens the distance: informational erosion. A component-level bill of materials exists at design stage, is diluted during manufacture through supplier substitutions, is commonly reduced in the as-built set to an assembly drawing, and finds no place in the file set that enters the data room when the asset changes hands. The question asked twenty years later is narrow and specific — which alloy is this profile, which chemistry is this coating, which adhesive class is this joint. Where the answer is not on record, the recovery facility operates on a contaminated-stream assumption and the material steps down a grade, moving from structural alloy toward casting alloy. That step down is irreversible, and its cause is documentary rather than physical.
The balance-sheet expression of all this sits in the decommissioning provision. The provision is typically constructed by netting expected scrap revenue at end of life against removal cost, and the more positive the scrap assumption is held, the smaller the provision becomes and the more comfortably the model closes. In configurations where separation labour exceeds scrap revenue, however, the net line changes sign, and the provision is then not merely undersized but constructed in the wrong direction. Stressing that line against separation hours rather than against commodity price is the single calibration that makes the underlying exposure visible, and it is rarely the sensitivity the model was built to run.
The second surface is the transaction table. In a portfolio transfer, what a buyer asks about decommissioning liability is not the estimate but the basis of the estimate; absent a component-level material record, the figure ceases to be a verifiable calculation and becomes a represented assumption, and a buyer prices an unverifiable assumption not at its expected value but at the upper end of its uncertainty band. That pricing emerges through one of three channels in practice — a headline discount, a specific indemnity, or an escrow tranche that is not released at closing. The same uncertainty reappears in prequalification for corporate offtakers requiring recycled-content undertakings, where it accumulates not on the cost side but as an eligibility condition on the revenue side.
A third surface becomes visible long before end of life, in the operating budget. A joint that makes a component inseparable in situ also makes it unrepairable in situ; a fault that would otherwise call for a part replacement escalates into a module replacement, and module replacement stretches spares policy, logistics windows and availability undertakings simultaneously. The circularity gap therefore reaches the O&M budget as spare-part cost and downtime well before it reaches the balance sheet as a provision, and in that earlier location it is usually tracked under a separate heading, unconnected to the joining decision that produced it. Seeing that the two lines share a single root is the precondition for anyone owning the problem at all.
This tendency is managed through decision architecture rather than individual awareness, and the architecture has four components. The first is the material record: alloy, coating, joining method and supplier declaration held at component level as a single record opened at design and carried through the as-built set into the transfer file. The second is the joining decision record: every irreversible joint written up at the moment of proposal rather than the moment of approval, together with its whole-of-life rationale, so that the rationale functions as a comparable input rather than a retrospective defence. The third is binding the end-of-life line in the investment model to separation hours rather than to scrap price. The fourth is contractual — the material declaration made a deliverable, the take-back undertaking a defined clause in the supply agreement, and documentation completeness a condition of a payment milestone.
BEIREK places this intervention inside the project management line rather than running it as a separate sustainability workstream. Ahead of specification freeze we operate a review in which irreversible joints are listed and each one's whole-of-life rationale is recorded alongside a designated counter-argument role; the output of that review is not an advisory note but a decision table forming part of the design record. The material record is written into the acceptance conditions for the as-built documentation set, supplier material declarations are tied to a payment milestone, and the integrity of that set within the transfer file is carried as a discrete item on the pre-closing checklist rather than absorbed into a general documentation heading.
On the financial side we recalibrate the end-of-life line on the same rhythm as capex contingency — within the quarterly cost review — and test the provision's sensitivity against separation labour and site access conditions rather than against commodity pricing. What that distinction means differs by party: for the sponsor it is the defensibility of a model assumption; for the EPC contractor it is coherence between the specification and the document set actually delivered; for the lender it is the weight the provision carries in the covenant calculation; and for a future buyer it determines whether what arrives at the table is a discount or a figure capable of being verified.
End-of-life cost is never created at end of life; it is created on the day the specification is frozen, and it then waits, unrecorded, for twenty years. The question worth asking of an asset is therefore not whether it is built from recyclable material, but whether it has been documented today what separating that material will cost, and to whom.