Cycle

What it is (precise)

A full cycle moves energy equal to the system's usable capacity out of, and back into, the battery. Real grid operation rarely runs 100% to 0% State of Charge and back, so engineers count Equivalent Full Cycles: partial charge and discharge events, summed by energy throughput and normalized to rated usable capacity. Two 50%-depth swings book one EFC. Irregular state-of-charge traces are scored either by simple throughput division or by rainflow counting — an algorithm borrowed from fatigue analysis that decomposes the trace into equivalent full and half cycles.

The two methods agree on steady daily duty and diverge on jagged profiles, which is exactly where the choice starts to matter. Warranty and degradation models are written in EFC and cumulative throughput (MWh), not in calendar starts and stops or the raw count of dispatch events. Whenever a vendor quotes cycle life, that number is denominated in EFC — learn to think in throughput first, event count never.

Pin down the reference capacity behind the count. Usable capacity is not nameplate — the BMS restricts the operating window — and it is not constant, because capacity fades over life. Most warranty frameworks normalize EFC to beginning-of-life usable energy at the DC terminals.

A counter normalized instead to the current, State of Health-adjusted capacity will report more cycles from the same throughput as the asset ages: a plant delivering a fixed daily MWh block registers a constant EFC rate against the beginning-of-life reference but a rising one against as-measured capacity. Cycling degradation also sits on top of calendar (time-based) fade, so cycle count alone never fully describes how far along the aging curve a system sits. Two identical EFC totals on differently aged assets are not the same wear.

Why it matters in a real grid-scale project

Cycle count is the link between how the asset is dispatched and how fast it loses capacity. Each EFC adds cycling degradation, which stacks on calendar fade. The supplier warranty caps annual throughput — a stated number of EFC per year, cycles per day, or a lifetime MWh limit — and guarantees a minimum retained capacity at end of term.

Exceeding the cycle budget can void the energy-retention guarantee, so the operator has to dispatch inside the warranty envelope. The EMS is usually configured to track cumulative EFC and throttle dispatch before the cap is breached. The cycle is where the trading desk meets the warranty desk.

The friction is that markets count events while warranties count energy. A day with a morning and an evening discharge reads as two cycles on the trading schedule even if both were shallow; under throughput accounting those two half-depth events sum to roughly one EFC. A price optimizer left unconstrained will happily run two full cycles a day when spreads are wide, consuming the annual budget twice as fast as the warranty assumed — so the trading strategy has to price the marginal cycle it is about to spend against warranty life, not just against tomorrow's spread.

The assumed cycle count drives augmentation planning. To hold a contracted POI energy delivery flat over a 15-20 year term against cycling and calendar fade, projects either oversize usable energy on day one or add battery Rack capacity in later years. The annual EFC assumption feeds the bankable performance model, the O&M and augmentation reserve, and the LCOS.

An LCOS number is meaningless without a stated cycling assumption: the same capital spread over 200 versus 365 cycles a year roughly doubles the cost per discharged MWh. Whenever you see an LCOS figure, the first question is: at how many cycles per year?

Typical values and standards

Lithium Iron Phosphate cells, the dominant stationary chemistry, are typically rated on the order of 6,000-10,000+ EFC to a ~70% retention threshold; some vendors quote more to a 60% threshold. NMC cells typically deliver materially fewer — roughly 3,000-5,000 EFC under comparable conditions — a large part of why LFP became the grid-storage default where energy density is not binding.

Sodium-ion vendors claim cycle life in LFP's range, but the utility-scale field record is still short. EV cells are designed for perhaps 1,000-3,000 deep cycles, an entirely different duty specification you should never benchmark a stationary asset against.

Duty profile sets the burn rate. Daily energy arbitrage books roughly one EFC per day, about 300-365 EFC per year; a typical 4-hour system does this at 0.25C. Frequency regulation produces hundreds of shallow swings a day, but their EFC total — summed by throughput — usually lands far below the event count.

Warranties commonly permit around one cycle per day (about 365 per year) or an equivalent annual MWh throughput; heavier duty of up to roughly two cycles per day is negotiated and priced separately. Realized cycle life depends strongly on Depth of Discharge, C-rate, cell temperature, and resting SOC. Deep, fast, hot cycling at high average State of Charge burns through the budget fastest, and all four levers appear in the datasheet fine print.

On the performance side, IEC 62933-2-1 covers unit parameters and test methods for electrical energy storage systems, and IEC 61427-2 defines on-grid duty-cycle test regimes; contractual capacity tests are built on such guidance. But no standard scores field usage for a warranty — how partial swings are decomposed, which capacity reference the division uses, and where the energy is metered are all set in the warranty annex. Separate cycle life from the safety and installation envelope too, and know each standard's role — they are routinely confused.

Cycle-life claims come from vendor lab protocols, often continuous cycling at 25 C and moderate C-rate, not from any certification. IEC 62619 and UL 1973 cover safety from cell to rack. UL 9540 certifies the ESS product. UL 9540A is the fire-propagation test method feeding NFPA 855, which governs installation. NFPA 68/69 cover deflagration protection. None of these certifies a cycle-life number — that lives only in the datasheet and the warranty, so never cite a safety mark as evidence of longevity.

How it shows up in specs, studies and contracts

On a cell or system datasheet, cycle life appears as a headline number plus retention curves: capacity versus EFC at stated conditions. Never take the headline alone. Read the Depth of Discharge behind it (100% or 80%?), the charge and discharge C-rate, the test temperature (usually 25 C), and the end-of-life threshold — a 60% threshold quotes far more cycles than 70% for the same cell.

The single question that reprices most bids: ask the vendor for retention curves at your actual duty profile and site temperature, plus the cycling model behind them, not just the standard lab protocol. Lab-to-site is exactly where optimistic cycle numbers collapse.

In contracts, the cycle appears as the warranty's throughput cap, a year-by-year capacity-retention table indexed to a stated cycles-per-year assumption, and an annual capacity test that verifies retention against it. Tolling and offtake agreements specify how many cycles per year the offtaker may call and at what depth.

Check six things: whether cycles are counted at the DC terminals, the PCS AC terminals, or at the POI (auxiliary loads and conversion losses make these differ), how partial cycles are aggregated, whether unused cycle budget rolls over year to year, what happens to the retention guarantee if the cap is exceeded in one year, which cell revision the retention curve was measured on, and whose meter and logs govern at claim time. Periodic capacity tests spend throughput too — a witnessed full charge-discharge test is by definition about one EFC — so test frequency is itself a negotiated number.

Common pitfalls

Lab cycle life does not transfer directly to the field. Vendor protocols cycle continuously at controlled temperature; a real site sees seasonal heat, partial cycles at varying depth, and long rests at high SOC that add calendar fade. Degradation is also not throughput-linear.

Two 50% swings book the same EFC as one 100% swing but generally cause less wear, while the same throughput delivered hot and fast causes more. EFC is the accounting unit, not a physics model. Bankable degradation models track depth, temperature, C-rate, and resting SOC alongside the cycle count, and you should plan for field life below the datasheet headline.

Comparing cycle counts across incompatible bases is the tender-stage version of the same error. A 10,000-cycle claim at 80% Depth of Discharge to a 60% retention threshold is not better than an 8,000-cycle claim at full depth to 70%: normalized by throughput, the first is about 8,000 EFC to a laxer end condition. Normalize depth, end-of-life threshold, and temperature before ranking bids.

Then guard against accounting drift in operations: an EMS that counts cycles by a different method, denominator, or metering point than the warranty annex produces a discrepancy nobody sees until a capacity claim is denied on logged throughput. Treat the cycle definition as an operating specification — metered, logged, and reconciled against the contract — not as vocabulary.

Do not let cycle life stand in for asset life. A lightly cycled asset — say a capacity-market battery running under 100 EFC per year — is dominated by calendar fade and hits its end-of-life threshold on the clock, not the odometer.

Quoting an 8,000-cycle cell as 22 years at one cycle per day makes the opposite mistake: calendar and cycling fade stack, and the warranty term (typically 10-20 years) usually expires before the cycle budget does. Always model both aging channels and check which one binds for your dispatch profile. That answer decides whether more cycles or a longer term is worth paying for.