Pre-COD degradation

What it is (precise)

The clock starts at the factory. A cell's capacity is measured after formation, and the factory acceptance test record is the beginning-of-life fingerprint the whole project will later argue from. Between that measurement and COD sit ocean freight, customs and inland haul, weeks to months in a site laydown yard, installation, energization, and a commissioning-plus-testing window that commonly runs 2 to 4 months on its own for a utility-scale plant.

Through all of it the cells deliver no energy; they sit. Sitting is exactly the regime Calendar aging describes — solid-electrolyte-interphase growth consuming cyclable lithium as a function of time, dwell SOC and temperature — so pre-COD degradation is not a separate mechanism but a separately owned slice of the calendar curve, the slice that runs before anyone is watching a revenue meter.

Two distinctions keep the accounting honest. First, self-discharge is not fade: a stored cell's open-circuit voltage and indicated charge drift down over months, and most of that recovers on recharge. Irreversible calendar fade does not recover, and only it belongs in the degradation ledger. Second, aging is not damage.

Cells fading along the vendor's own storage curve are behaving exactly as modelled; cells pulled below their minimum voltage because BMS and monitoring electronics kept drawing from an unattended rack are damaged, and a warranty treats the two very differently — the first is allocated, the second is usually excluded and argued over.

The position on the curve is what gives the window its weight. Calendar fade follows an approximately square-root-of-time trajectory — steepest at the start — with modern LFP cells showing on the order of 2-4% total fade in the first year under moderate SOC and temperature.

A project that takes about a year from cell manufacture to COD therefore arrives at its capacity test a couple of percent below the factory measurement with nothing wrong at all. And the rate is thermally activated, roughly doubling per ~10 C rise in cell temperature, so containers delivered at a high SOC and parked through a summer in an uncooled yard age materially faster than the curve the proposal assumed.

Why it matters in a real grid-scale project

The first casualty is the acceptance test. COD capacity tests commonly require demonstrating roughly 95 to 100 percent of the contracted MWh at the point of interconnection, and a guarantee referenced to factory nameplate can fail that test on aging that happened precisely on schedule — the contract simply never assigned the transit and storage months to anyone.

The fix is contractual, not technical: state the guaranteed capacity at the test date, with pre-COD fade explicitly allocated, or oversize enough at day one that the steep early fade is absorbed. Because the COD test result then serves as the baseline every later warranty dispute references, the handling of pre-COD fade decides where a 15-20 year degradation argument starts from.

The second casualty is warranty life. Warranty periods can be defined to start at delivery, at energization or at COD, and the difference is not cosmetic: on a project where the interconnection, the transformer or the permit slips, cells sit in storage while a delivery-referenced warranty burns through its early years with zero revenue against them.

Risk allocation follows possession — the supplier typically owns the cells through transit until delivery, the owner owns the laydown yard and construction storage — so a schedule delay quietly converts into a degradation and warranty-headroom cost that lands on whoever holds the cells when the clock runs. Logistics decisions are degradation decisions in this window, which is why supply agreements mandate a shipping SOC, commonly around 30-50%, and recharge intervals if energization slips.

Typical values and practices

The working numbers are the calendar-aging numbers, applied to the front of the curve. LFP first-year total fade on the order of 2-4%, tapering toward roughly 1-2% per year; fade rate roughly doubling per ~10 C rise in cell temperature; high dwell SOC accelerating fade, which is one reason shipment at a moderate 30-50% SOC is the mandated norm rather than a courtesy.

All of these are scenario-specific — a curve quoted at 25 C and moderate SOC says nothing about a container at 90% SOC in a 40 C yard — so every pre-COD estimate should carry its assumed storage temperature and SOC alongside the percentage.

The binding text is not a standard but the vendor's storage requirements, usually incorporated into the warranty by reference. A typical battery storage specification states an allowable storage temperature band, a maximum storage duration before a mandatory recharge, and the SOC window to recharge into; exceeding any of them is a warranty condition, not advice.

Good procurement practice adds a cap on cell age at delivery, manufacture dates on the shipping documentation so the age is checkable, and temperature loggers riding in the containers so storage conditions are provable later. No safety or performance standard sets pre-COD degradation limits — the supply agreement and the storage manual are the whole rulebook.

How it shows up in specs, studies and contracts

In the battery supply agreement, look for four things: the warranty start definition (delivery, energization or COD — establish which, and what happens to each on a slipped schedule); the reference for the guaranteed capacity (factory nameplate versus capacity at the test date); the storage obligations and who performs them; and the evidence requirements.

The storage obligations hide a practical trap — recharging stored containers requires a power source, and on a construction site the PCS and auxiliary supply may not be energized yet, so a recharge obligation without a named power supply and scope owner is an obligation nobody can discharge. The division-of-responsibility matrix should say who provides temporary power, who executes the recharge, and who logs it.

In the financial model and the independent engineer's review, the question is where year zero sits. Warranty degradation tables count anniversary years from their defined start date, but the vendor's underlying fade curve started at manufacture; if the table's year zero is COD while the curve's year zero is the factory, the offset is either absorbed in the vendor's margins or silently missing from the model.

Independent engineers check exactly this alignment, along with whether the COD test conditions — temperature, resting SOC, measurement point — were recorded well enough for the baseline to survive a dispute a decade later. A weak or unallocated pre-COD assumption shows up years on, as a capacity claim that fails because nobody can prove what the battery had when commercial life formally began.

Common pitfalls

The classic failure is the nameplate-referenced acceptance test: the contract guarantees a percentage of factory nameplate at COD, the project takes a year to get there, and the test comes in short by roughly the calendar fade the vendor's own curve predicted. Nothing malfunctioned — the drafting did.

Its twin is the delivery-referenced warranty on a slipping project, where each month of interconnection delay consumes warranty life with no revenue against it. Both are cheap to fix at signing and expensive to argue afterwards, which is why cell-age caps, test-date-referenced guarantees and explicit allocation of the storage window belong in the term sheet, not the punch list.

The operational failures are storage failures. A missed recharge lets the parasitic draw of rack electronics pull cells below their minimum voltage — over-discharge damage that warranties typically exclude, categorically worse than the aging it was supposed to prevent.

A hot laydown yard, or containers left at a high SOC because nobody owned the discharge-to-storage-SOC step, multiplies the fade rate the model assumed away. And the evidence burden falls on whoever stored the cells: without temperature and SOC logs from the storage period, storage compliance cannot be proven, and a later capacity claim can die on that gap alone. Log it as if the warranty depends on it, because it does.