Degradation

What it is (precise)

All degradation stems from a short list of physical and chemical processes inside the cell — chiefly growth of the solid-electrolyte interphase (SEI) that consumes cyclable lithium, loss of active material in the electrodes, and lithium plating when cells are charged fast or cold — expressed as two measurable trends: stored energy goes down, and internal resistance goes up.

For the LFP chemistry that dominates stationary BESS, fade is gradual and fairly predictable across most of life; NMC, used more rarely in stationary projects, tends to fade faster and is more thermally sensitive. The practical levers an operator controls are temperature, average state-of-charge, depth of discharge, C-rate, and annual cycle count.

The two aging channels superpose. Calendar aging runs on time, temperature and resting SOC even in a parked system; cycle aging runs on throughput, depth of discharge and C-rate. Which dominates depends on duty — a daily-cycled arbitrage asset is cycle-dominated, a rarely dispatched capacity resource is calendar-dominated.

Both are quantified against a beginning-of-life reference: retained capacity in percent of BOL energy, fade rate in percent per year, and cumulative duty in equivalent full cycles (EFC) or MWh of throughput. The Calendar aging and Cycle aging entries cover the mechanisms and their accelerants in depth; this entry is about what degradation does to the system and its contracts.

The reference point matters as much as the number. Cell-level fade measured in a laboratory is not the same as system-level fade measured in MWh discharged at the AC terminals, and neither equals energy delivered at the point of interconnection after transformer and auxiliary losses. A serious specification always states the basis: DC or AC, cell or system, BOL or a dated capacity test, and the exact definition of an equivalent full cycle used to count throughput.

The four fades

Capacity fade is the shrinkage of the energy the cells can store, measured in DC MWh at the racks against the BOL rating: less cyclable lithium, shorter duration at rated power, and a year-by-year retention table in the capacity warranty to enforce it. Usable-energy loss is the system-level version the offtaker actually experiences at the POI, and it runs ahead of the average cell fade: series strings are capped by their weakest elements, so rack-to-rack imbalance, the BMS operating window and auxiliary consumption all sit on top of cell-level loss.

Its home is the contracted usable AC energy in the tolling or offtake agreement and the capacity-test procedure that settles disputes — and because only the POI number pays, the gap between rack MWh and delivered MWh is a recurring source of warranty argument.

Power fade is resistance growth read as a power limit: a thickening SEI and degrading interfaces make the cell sag further under load, so it hits its voltage limits earlier at high C-rate, at low temperature and near the ends of the SOC window — and a cell can fade in power faster than in energy.

It is the least-contracted of the four, because End of Life and warranty tables are almost always written against energy, so a plant can keep passing its capacity test while quietly losing the ability to hold rated MW across the full discharge — including power owed at the POI under grid-code obligations late in life.

Efficiency fade is the same resistance growth read in the energy account: I²R losses grow with age, so round-trip efficiency declines from its commissioning value — typically in the 85–90% AC-AC range — and every lost point is throughput bought and not sold; its home is the RTE guarantee in the supply or service agreement, tested at its own stated boundary. The four fades share mechanisms and are correlated, but they are not interchangeable: a system can meet its capacity guarantee while missing its RTE guarantee, and vice versa, which is why each carries its own test and its own remedy.

Why it matters in a real grid-scale project

Degradation is a commercial problem before it is a chemistry problem. A BESS is sold to deliver a guaranteed energy — say a contracted 400 MWh over four hours — across a 15–20 year term. Because capacity fades, the plant must be oversized on day one, or capacity must be added later, so the aged system still meets its End of Life energy target.

This oversizing-versus-augmentation trade is a central line item in project finance: day-one overbuild of 10–25% above the contract quantity is common, cumulative augmentation often adds 15–40% of the original DC build over the term, and the chosen strategy directly sets the bankability of the asset and its levelized cost of storage.

The capacity warranty is where degradation is codified. Suppliers guarantee a minimum retained energy — usually a year-by-year retention table or curve — provided the owner stays inside an operating envelope: cycles or throughput per year, maximum C-rate, cell temperature limits enforced by the thermal-management system, and SOC dwell rules.

Violating the envelope voids the warranty regardless of measured health, so EMS dispatch logic and the trading strategy must treat those caps as hard constraints, and the augmentation plan is engineered around the degradation model. Resistance growth matters in parallel: it raises auxiliary and cooling load and drags efficiency and deliverable power down over life, as the four fades above describe.

Typical values and standards

Useful life is conventionally tracked to a retained-capacity floor, commonly around 65-70% of BOL energy, after which the system is at End of Life for the application — the cells are not dead, just below contract. Modern LFP systems are typically warranted for 6,000–10,000+ equivalent full cycles (some vendors quote more at reduced depth of discharge), supporting one to two cycles per day over 15–20 years; NMC generally tolerates roughly 3,000–5,000 under comparable conditions.

First-year fade is usually the largest — often 2–4% — before settling to a near-linear 1–2% per year. A common rule of thumb is that calendar aging roughly doubles for every 10°C rise in cell temperature, and dwelling at high SOC accelerates it further; that is why fleets are parked mid-SOC and cooled tightly.

No standard dictates how fast a battery may fade — degradation limits are contractual. The standards define how to measure and how to stay safe. IEC 62933-2-1 provides unit parameters and test methods for electrical energy storage systems, IEC 62620 covers performance testing of industrial lithium cells, IEC 61427-2 addresses on-grid duty-cycle test regimes, and supplier capacity-test procedures verify retention against the warranty.

The safety framework interacts with aging: NFPA 855 governs installation, UL 9540 is the ESS product safety listing, UL 9540A is the thermal-runaway fire-propagation test method whose data informs spacing and deflagration protection (NFPA 68/69), and UL 1973 and IEC 62619 both cover cells and packs. Aged, higher-impedance cells run hotter, which is why degradation, thermal management, and fire safety are evaluated together.

How it shows up in specs, studies and contracts

On a cell or DC-block datasheet, degradation appears as cycle-life curves — retained capacity versus cycle count at stated temperature, C-rate, and depth of discharge. Check the reference conditions: a 10,000-cycle claim (always to a stated SOH threshold, e.g. 70-80%) at 25°C and 0.5C does not transfer to a hot site cycling harder, and claims quoted to different end thresholds are not comparable until normalized — 10,000 cycles to a 70% floor is a weaker claim than 8,000 to 80%, not a stronger one.

In the supply contract it appears as the capacity warranty table (guaranteed percent retention per year), the operating envelope that conditions it, and the EFC or MWh-throughput definition used for counting. Confirm whether the guarantee is at DC terminals or AC, at what temperature reference, and how augmentation energy is treated in the count.

And because warranties are written against energy, power fade goes unwatched by default: a project that needs rated MW for the full duration late in life should write a duration-at-rated-power or power-capability test into the acceptance and warranty regime explicitly.

In the financial model, the degradation curve drives the augmentation schedule, the energy available for revenue in each year, and the LCOS — which is why lenders' independent engineers scrutinize it and benchmark it against cycle-test data and third-party degradation models. In operations it is verified by periodic capacity tests, typically annually or at warranty milestones, and disputes usually reduce to test method and reference conditions.

Questions worth asking on any project: what is the guaranteed retention at year 10 and year 20, what cycling profile is assumed, is the curve a floor or an expected value, how is availability treated separately from capacity, and who pays for augmentation if the fleet underperforms.

Common pitfalls

The most common error is treating the warranted curve as a forecast. It is a floor the supplier is confident of clearing; the expected fade is usually better, and the financial model should carry both cases. A close second is collapsing the four fades into one number: capacity fade shrinks the tank while efficiency fade taxes every cycle through it, and a healthy RTE says nothing about retained energy.

A related trap is mixing bases — cell-level lab fade, system-level AC fade, and POI-delivered energy are three different numbers, and quoting one against another creates phantom margin. Also distinguish degradation from availability: a tripped container loses revenue but recovers; degraded capacity does not.

Watch for nonlinearity and masking. Fade curves can show a "knee" where loss accelerates late in life, particularly under high temperature, high average SOC, or sustained lithium plating, so extrapolating early near-linear data to year 20 is optimistic.

The BMS presents SOC over a managed usable window, and window re-mapping can hide underlying fade from casual observation while absolute capacity keeps shrinking; container-average temperatures likewise hide hot racks that age faster and drift out of balance. And the BMS-reported state of health is an estimate, not a measurement — contractual capacity is established by a discharge test under defined conditions, and drift between the estimate and the tested value is a recurring source of warranty friction.