Duty profile

What it is (precise)

Strictly, the duty profile is the full time series of power and state of charge the cells experience. Nobody writes contracts against a time series, so practice reduces it to a handful of parameters: cycle count (in Equivalent Full Cycles per year, or cycles per day, or annual MWh throughput — the accounting is covered under Cycle), the Depth of Discharge of those cycles, the charge and discharge C-rate, the resting or Average State of Charge between dispatches, and cell temperature.

Those five axes are the inputs every serious degradation model takes, because they drive the two aging channels: throughput, depth, and rate feed Cycle aging, while time, temperature, and average SOC feed Calendar aging. A duty profile is the joint specification, and the axes are not interchangeable — the same annual EFC delivered as shallow, cool, mid-SOC cycling and as deep, hot cycling with long full-charge rests are different duties with different fade trajectories, even though a throughput counter cannot tell them apart.

The duty profile exists in three written forms, and keeping them distinct is most of the discipline. The reference duty is what the vendor's cycle-life data was measured at — typically continuous lab cycling at 25 °C and moderate C-rate, without the rests, partial cycles, and seasonal temperature swings a site produces.

The contracted duty is the operating envelope in the warranty annex and the dispatch-rights definition in the tolling or offtake agreement: the cycles per year, DoD, temperature window, and resting-SOC conditions the guarantee is valid inside. The as-operated duty is what the BMS and EMS logs say actually happened. Degradation predictions are made on the first, priced on the second, and judged on the third.

Why it matters in a real grid-scale project

Each revenue service has a recognizable duty signature, and choosing the service chooses the wear. Daily energy arbitrage books roughly one EFC per day at or near full usable depth — about 0.25C on a 4-hour system — with the fleet swinging through the whole SOC window. Frequency regulation produces hundreds of shallow swings a day whose throughput sums to far less than the event count suggests, usually run from a mid-SOC resting point that keeps headroom in both directions.

A capacity or reserve product is the opposite duty: very few cycles, long idle periods, and a commercial pull toward parking near full charge for readiness — which makes it calendar-dominated, since calendar fade is what binds for assets running under ~100 EFC per year, and high resting SOC with temperature is exactly what accelerates it. A lightly cycled battery is not automatically a lightly aged battery.

The duty profile is also the load-bearing assumption of the financial model. The warranted retention curve in the Capacity warranty holds only for the contracted duty; the augmentation schedule is read off that curve, so its tranche timing inherits the same assumption; and an LCOS figure is meaningless without the cycling assumption stated next to it.

This is why reconciling actual dispatch against the warranted envelope is a standing operational task, not a design-time check: the profile the market rewards drifts away from the profile the warranty assumed, and the mismatch surfaces years later as a capacity test below the guaranteed table. Reconciliation means reading the logged duty — depth, rate, temperature, resting SOC, cycle count — against the envelope axis by axis, not glancing at one counter.

Revenue stacking multiplies the problem, because the real duty is a blend that no single service's reference data describes. A plant that runs regulation through the day and an arbitrage cycle in the evening produces a jagged SOC trace whose wear is not the sum of two clean profiles, and whose cycle accounting depends on the counting method the warranty annex picked.

Bankable projects handle this by making the vendor state the retention curve at the project's actual intended duty profile and site temperature — not the lab protocol — and by keeping the modelled duty, the contracted duty, and the dispatch strategy consistent with each other at financial close.

Typical values and standards

Contracted duties cluster tightly. Warranty envelopes commonly permit around one cycle per day — roughly 300-365 EFC per year at a defined Depth of Discharge — with heavier duty of up to roughly 1.5-2 cycles per day negotiated and priced separately. Modern LFP systems are routinely warrantied for daily cycling at or near full usable depth.

Under a normal utility duty inside that envelope, LFP fleets typically fade about 1.5-3% per year — with a front-loaded first-year step of around 2-4%, then roughly 1-2% per year — toward a warranted End of Life retention commonly near 65-70% of beginning-of-life capacity over a 15-20 year term. Every one of those figures is conditional: run hotter, deeper, faster, or parked higher than the stated duty and the fade outruns the table.

On the standards side, IEC 61427-2 defines on-grid duty-cycle test regimes — standardized profiles for services like frequency regulation and time shifting — and IEC 62933-2-1 covers unit parameters and test methods for electrical energy storage systems; contractual capacity and efficiency tests are built on such guidance, each valid at its own stated duty point.

But no standard scores a field duty profile against a warranty. How the as-operated duty is reconstructed, which parameters are checked, and what an excursion costs are all set in the warranty annex, which makes the duty profile a negotiated contract object wearing the clothes of an engineering quantity.

How it shows up in specs, studies and contracts

On a datasheet the duty profile hides in the conditions under the cycle-life headline: the DoD, C-rate, and temperature of the test protocol, and the end-of-life threshold the count runs to. In the battery supply agreement it becomes the operating envelope — cycles or throughput per year, DoD, temperature window, resting and average SOC conditions — that the Capacity warranty is valid inside.

In a tolling agreement it appears as the offtaker's dispatch rights, and the two documents must be checked back-to-back: a toll that permits more annual cycles than the warranty allows leaves the owner silently carrying the gap — a toll permitting 400 cycles per year over a warranty that voids above 300 is the classic form. In the financial model it is the cycling assumption behind the revenue stack, the augmentation reserve, and the LCOS.

Operationally, the duty profile is an evidence trail. BMS and EMS logs — throughput, SOC histograms, temperature history, C-rate excursions — are what a capacity claim is adjudicated against, so the logging and retention regime is itself a contract term worth reading.

Three checks earn their keep on any project: whether the duty assumed in the revenue model, the duty permitted by the warranty, and the duty granted to the dispatcher are the same numbers; whether the vendor's retention curve was restated for the project's duty and site temperature rather than quoted from the lab protocol; and who pays — for augmentation, or in lost warranty cover — if merchant dispatch runs the plant harder than the plan.

When reality diverges from the assumed duty

Over a 15-20 year term the market that set the original duty assumption will not survive intact. Regulation markets saturate as storage enters, arbitrage spreads move with the generation mix, and new products appear — so the year-1 duty profile is a snapshot, not a plan. When dispatch drifts heavier than assumed, the consequences arrive in order: real fade outruns the warranted curve, the augmentation tranches move earlier than the reserve was sized for, and if the drift breached the envelope, the supplier can void or pro-rate exactly when the guarantee is needed.

Drift can be silent: an EMS strategy change that parks the fleet at higher SOC between dispatches alters the duty profile without changing the cycle count at all, and nothing flags it until a claim fails the envelope check. The defense is to treat the contracted duty as an operating specification — logged, alarmed, and reviewed — and to re-forecast degradation against the measured duty at each capacity test rather than trusting the financial-close curve for two decades.

The divergence is asymmetric. Running heavier than assumed burns warranty budget and can void cover, but running lighter does not refund life: an underused asset slides toward the calendar-dominated regime, where time, temperature, and resting SOC set the fade and the unused cycle budget expires with the warranty term.

The practical posture is to treat the duty profile as a live operating variable with a price on every axis — the marginal cycle, the extra depth, the hours parked full each carry a cost in warranty life — and to make dispatch decisions against those prices, not only against tomorrow's market spread.