Decommissioning plan

A permit condition, not a courtesy

The plan is usually demanded before the project is built. Zoning approvals and conditional-use permits for grid-scale storage commonly require a decommissioning plan with the application or as a condition of approval, and where the site is leased the removal and restoration obligations sit in the land document as covenants attached to the land rather than to whoever happens to own the project.

NFPA 855 adds its own hook in US practice, carrying decommissioning provisions alongside the commissioning ones. The practical effect is that the plan is written by development-stage people, for work that will be done by people who have not been hired, at prices nobody can see, under rules that in part do not exist yet.

Naming the responsible party is the clause most often written loosely. The obligation belongs to the project company, but a grid-scale asset changes hands — developer to sponsor, sponsor to an infrastructure fund — and the operating contracts change with it. What keeps the obligation attached is that the permit condition binds the land use and the lease covenant binds the land; what makes it collectable is the financial assurance posted alongside.

The plan should say who executes the work, who signs off that it is complete, and what evidence closes the obligation: a survey, disposal receipts, recycling certificates for the batteries. "Restored to the landowner's satisfaction" is not a completion criterion anyone can act on two decades later.

Then the update clause. A plan approved today describes a job in the 2040s using today's recycling industry and today's law, and both will have moved — EU recycling-efficiency and material-recovery targets step up on fixed dates through 2031, storage-specific rules at US state and county level are still being written, and the processing capacity that will take the black mass may not have been built yet.

A plan with a review cycle, where the estimate is re-run and the applicable rules re-checked and the document re-filed, stays useful. One that is filed once and never reopened becomes a description of an industry that no longer exists.

Discharging a plant that cannot be switched off

A battery is not de-energised by opening a breaker. Every rack holds its stored energy until that energy is taken out of it, and each string sits at its own terminal voltage regardless of what the switchgear is doing — on a 1500 V-class DC architecture that is lethal potential inside an enclosure that has been shut down for a week. There are two ways out: through the power conversion system, exporting to the grid, or into load banks that dump the energy as heat.

Both take real time on a large site, and the first carries a scheduling consequence people miss. Exporting the last of the stored energy needs the interconnection still live, the converters still working and someone still contracted to operate them. Decommissioning has to be sequenced before the interconnection agreement is terminated and before the O&M contract runs out, not after.

Inside the fence the sequence runs grid-side inward and only then down to rack level. Disconnect at the point of interconnection under the utility's own procedure, isolate and earth the main power transformer, work back through the medium-voltage switchgear and the converter blocks with lock-out and tag-out at each stage.

Rack-level isolation comes last: open rack breakers, pull string fuses or connectors, and treat every module as live until proved otherwise. Discharge to a low state of charge, not to zero — deep over-discharge damages cells and can plate copper from the current collector, turning a stable module into a hazard during handling. What counts as a safe transport state is set by the applicable packing instruction and the carrier, not by one universal number.

A module that has been through a thermal event does not join this process at all. Damaged, defective or suspect cells and modules have their own handling regime: quarantined on site, separated from the healthy stock, and moved under transport rules written specifically for them. Assume from the first draft of the plan that some fraction of the fleet will be in that category, because twenty years of operation on any site produces a store of failed modules that were pulled and never shipped.

Moving batteries is a dangerous-goods operation

Lithium-ion batteries are Class 9 dangerous goods, and which entry applies depends on how they travel. UN 3480 covers cells and batteries shipped on their own, UN 3481 covers them packed with or contained in equipment, and UN 3536 covers lithium batteries installed in a cargo transport unit — the entry that matters when a whole enclosure moves as one piece.

The obligations come from the modal regimes built on the UN Model Regulations: ADR and RID for European road and rail, IMDG at sea, 49 CFR 173.185 in the US. Cells and batteries have to have passed the UN 38.3 transport tests, which for equipment installed twenty years earlier means locating the original test summary rather than re-running anything.

Damaged and defective units are a separate regime again. Special provision SP 376 governs them, with dedicated packing instructions — P908 and LP904 for damaged or defective units generally, and P911 and LP906 for those liable to disassemble rapidly or react violently. Air transport of damaged or defective lithium batteries is barred outright.

Even healthy stock is constrained by mode: air freight caps standalone lithium-ion under UN 3480 at 30% state of charge, while road and sea carry no equivalent state-of-charge cap. Grid-scale removals therefore move by truck and ship, and an air-freight assumption anywhere in the schedule is usually a mistake.

Physical scale shapes the plan more than the paperwork does. Current liquid-cooled 5 MWh-class enclosures commonly ship at roughly 35-45 tonnes on a plan close to the 20-ft ISO footprint of about 6.06 m by 2.44 m, so crane selection, road permits and hardstanding are decommissioning inputs exactly as they were construction inputs — and the access roads have to still carry that load at the end of the asset's life, which is a question about the road rather than about the truck.

The plan needs a named receiving facility, because a truckload of Class 9 modules with no consignee is not a shipment. It also needs a trained shipper, since dangerous-goods declarations are signed by a person holding current training. That is a role to staff, not a box to tick.

Then the electrical plant, then the ground

With the batteries gone, what remains is conventional demolition carrying three material-handling problems. Liquid-filled transformers — the unit transformers and the main power transformer — hold mineral oil or ester fluid that has to be drained, contained and disposed of, and containment built for an unplanned spill now has to work for a planned one.

Medium-voltage switchgear may be SF6-insulated, and SF6 is a fluorinated greenhouse gas: the EU F-gas regime requires recovery by certified personnel at end of life rather than venting, so a plan that shows sealed switchgear going straight to a scrap yard is not a compliant plan. Cable is the third. Copper in trenches and duct banks is the most valuable metal left on the site, and whether it is recovered or abandoned in place is a permit and lease question before it is an economic one.

Civil restoration is where the land document takes over from the engineering. Foundations, plinths, duct banks and pull boxes come out to a depth the lease or the permit condition states; roads, hardstanding and fencing come out or stay according to what the landowner asked for at option signing; then regrading, topsoil replacement and revegetation against whatever restoration standard was written, with return to agricultural condition a common one that can carry a soil-compaction test with it.

A 100 MW / roughly 400 MWh four-hour US project has typically occupied on the order of 5 to 15 acres once fire-code aisles, access roads and setbacks are counted, and all of it is in scope. So is the gen-tie, unless the line was transferred to the utility — easements and wayleaves carry their own restoration obligations, and a plan that stops at the fence leaves the line, the structures and possibly the switchyard unaddressed.

Salvage, and what second life actually is

The plan almost always credits scrap value against cost, and the credits worth relying on are the ordinary metals: copper from cable and windings, steel from enclosures and structures, aluminium from busbar and conductor. Those have deep markets and a price you can look up on the day. The batteries are the opposite.

They are most of the mass, most of the labour and most of the transport cost, and their recovery value depends on a processing industry that is still forming. LFP dominates stationary storage and contains no cobalt and no nickel — the two metals whose value pays for NMC recycling. What is left to recover is lithium, copper, aluminium and steel, which is why LFP recycling has generally needed regulation or a gate fee to move rather than paying for itself.

That makes the salvage line a forecast rather than an estimate. Lithium prices have moved by large multiples inside single years, recycling capacity and gate fees are still being built out, and the EU is legislating recovery upward on a fixed schedule: Annex XII of Regulation (EU) 2023/1542 requires lithium-based batteries to reach 65% recycling efficiency by average weight from 31 December 2025 and 70% from 31 December 2030, with lithium material recovery at 50% from 31 December 2027 and 80% from 31 December 2031.

Rising mandated recovery is good for the material and neutral-to-negative for the estimate, because it raises the cost of compliant treatment. Carry the gross number as well as the net one in the plan, and say which assumptions the netting rests on, because the financial assurance argument is conducted on exactly that difference.

Second life belongs here as an option, not as an assumption. Under the EU Battery Regulation a battery that is repurposed is treated as a new battery being placed on the market, and the operator that repurposed it takes on the producer's conformity duties — testing, documentation and CE marking. The regulation also requires the battery management system of a stationary storage system to hold state-of-health and expected-lifetime data and to make it readable by the owner or by someone acting on the owner's behalf, which is what makes screening for reuse practical at all.

What does not follow the battery is the warranty. The original capacity guarantee was written for the original owner, the original duty and the original site, and repurposing ends it. A plan that assumes a buyer for twenty-year-old modules has assumed a market, a certification path, and a counterparty willing to take unwarranted cells.