Performance

Performance Guarantee

A performance guarantee is the set of contractual promises about what a grid-scale battery will measurably do — retain capacity against a year-by-year degradation curve, deliver a stated round-trip efficiency, and be available for dispatch a guaranteed fraction of the time — each backed by a defined test method and a remedy when the number is missed.

Performance guarantees are not a standalone document: they live inside the supply and service contracts, initially in the Battery Energy Supply Agreement (BESA, also written BSA) and for the operating years in the long-term service agreement, with the split varying by package. Every one of them has the same five moving parts — boundary, test, conditions, exclusions, remedy — and they are the clauses that turn a vendor datasheet into something a lender can underwrite.

Reviewed August 2026 by Sergey Syrvachev

New to BESS? Start free with the 7-email fundamentals course — no cost, no account.

The three numbers that get guaranteed

Almost every BESS performance package reduces to three families. Retained capacity: a capacity warranty commits the supplier to a minimum usable energy by project year, conditional on an operating envelope. Round-trip efficiency: a guaranteed RTE at defined test conditions, because efficiency losses are bought energy the model has already spent.

Availability: a guaranteed fraction of hours the system can respond to dispatch, with the measurement rules — planned outages, grid causes, force majeure — doing most of the negotiating work. Some packages add response-time or ramp commitments where the revenue depends on them, but capacity, efficiency and availability carry the money.

Five parts: boundary, test, conditions, exclusions, remedy

Every enforceable guarantee has the same five parts, and one that is missing a part is a dispute with a date on it. The boundary says where the quantity is measured — AC at the point of interconnection or DC at the battery terminals, with auxiliary load inside or outside the fence — and the same physical plant produces different numbers on either side of it.

The test says how it is measured: procedure, instrument accuracy, the state the plant must be in when the test starts, who witnesses it, how often it repeats, and how the raw result is normalised back to reference conditions. The conditions say what the plant must have been doing for the promise to stay alive, which is the operating envelope. The exclusions say what comes out of the calculation and on whose say-so. The remedy says what happens when the number is missed, and where the ceiling on that sits. Write those five for each guarantee and most arguments never start.

A guarantee without a test method is an opinion. Each promise carries a measurement protocol — a capacity test at defined temperature, state-of-charge window and C-rate; an efficiency measurement over a stated cycle; availability computed from agreed event categories — and the protocol is where disputes are won or lost long before anyone argues about money.

Every guarantee is also conditional on an operating envelope: cycle the battery harder, hotter or deeper than the envelope allows and the numbers void. That coupling runs both ways — the dispatch strategy earns the revenue, the envelope protects the guarantee, and a project that optimises one while ignoring the other has quietly un-guaranteed itself.

Remedies scale with the miss: liquidated damages or service credits for shortfalls, repair-or-replace obligations, and termination rights at the extreme. The credit behind the remedy matters as much as the remedy — a generous guarantee from a thinly capitalised entity is worth what that entity is worth, which is the same lesson bankability teaches about every promise in the stack.

A warranty promises the equipment is free of defects; a guarantee promises measured output with money attached — a battery can be defect-free and still miss its numbers.
BOUNDARYAC at the POI orDC at theterminals — thewhole conversionchain sits betweenthemTESTa named method,normalised toreferencetemperature andrateCONDITIONSthe operatingenvelope:throughput,temperature, SoC,C-rateEXCLUSIONSthe defined listREMEDYliquidated damagesor credits —bounded by its capand the obligor'screditRemove any one link and the number stops being enforceable in practice.

The guaranteed trio is retained capacity, round-trip efficiency and availability. Check the guarantee survives an assignment, an OEM change of control or a new O&M contractor — and that sole-remedy language does not hand the residual shortfall back to the owner.

Key facts
The guaranteed trio
Retained capacity (degradation curve), round-trip efficiency, availability
Five-part anatomy
Boundary, test, conditions, exclusions, remedy — a guarantee missing any one is unenforceable in practice
Boundary decides the number
AC at the POI including auxiliaries vs DC at the terminals is the whole conversion chain — state it for every guarantee
Normalisation
Measured results must be corrected to reference temperature and rate before comparison; an unwritten method yields two defensible answers
Where they live
BESA / BSA at delivery; increasingly the LTSA for the operating years; occasionally a standalone capacity-maintenance agreement
Always conditional
An operating envelope — throughput, temperature, SoC, C-rate — outside which the numbers void
Commissioning vs operating year
The reliability run gates COD; the operating-year availability guarantee is a separate number proven by a separate method
Remedies
Liquidated damages or service credits, repair obligations, termination at the extreme
What a remedy is worth
Bounded by its cap and by the obligor's credit; sole-remedy language hands the residual shortfall back to the owner
Must survive transfer
Check the guarantee follows an assignment, an OEM change of control, or a new O&M contractor

The five parts applied to each of the three

Retained capacity. Boundary: the contract must say DC terminals or AC at the POI, because the difference is the entire conversion chain. Test: an energy capacity test — a controlled discharge across the contracted SOC window at a defined rate and temperature — run on a stated cadence, with the beginning-of-life baseline established by the same procedure at commissioning. Conditions: the duty envelope and the throughput cap, evidenced from BMS and EMS logs rather than from anyone's recollection.

Exclusions here are rarely hours and usually causes — whether capacity sitting idle in modules pulled for warranty repair counts against the guaranteed curve is a drafting choice, not a physical fact. Remedy: repair, replace, augment or pay, in a stated order and against a cap. The step most often left vague is normalisation: a measured result has to be corrected back to reference temperature and rate before it can be compared with a table, and where that method is unwritten both parties will produce defensible numbers that disagree.

Round-trip efficiency. The boundary does more work here than anywhere else — AC-to-AC at the POI including auxiliary load sits several points below DC-to-DC at the battery terminals, and both are honest answers to different questions. Test: a defined cycle with stated start and end states of charge, a rest period, a specified power level, and an explicit treatment of auxiliaries, running as they would in service or excluded.

Conditions: temperature and power level, because efficiency is flat across neither. Exclusions: energy consumed by scope that is not the supplier's. Remedy: liquidated damages priced from the energy the missing points cost across the guarantee period, which is why an RTE shortfall is settled in cash far more often than it is repaired.

Availability. Boundary: the ability to deliver contracted MW and MWh at the POI, so a tripped HVAC skid or an MV breaker counts against the number even when every cell is healthy. Test: not a test at all but a calculation from the historian, which is why the formula, the data source and its resolution are the negotiation.

Conditions and exclusions carry more weight here than in the other two families — the exclusions annex is where availability guarantees are won and lost — and the choice between time-based and capacity-weighted accounting decides whether a partial derate reads as nothing or as a pro-rata loss. Remedy: a liquidated-damages schedule, commonly capped as a share of the annual service fee. Keep the commissioning gate separate: the reliability run that gates COD has its own threshold and its own reset rules, and passing it proves nothing about year seven.

Where they live: BESA, LTSA, or a capacity-maintenance agreement

The guarantees split across two contracts — occasionally a third, a standalone capacity-maintenance agreement — and the split is a negotiation, not a standard. Delivery-stage numbers — capacity and RTE demonstrated at commissioning — sit in the BESA or BSA, proven by the acceptance tests that gate final payment.

The long-tail numbers — the degradation curve, ongoing RTE, availability year after year — increasingly sit in the long-term service agreement, because a multi-decade promise is only worth much when the party making it is still on site maintaining the machine. Where a package puts each promise changes who owes the remedy and under which contract’s caps and termination rights — which is why the two documents are negotiated as one set.

The occasional third document is worth understanding on its own terms: a capacity-maintenance agreement commits a supplier to hold delivered energy at a contracted level by adding capacity when the curve calls for it, priced separately from the service fee.

It exists because guaranteeing a declining curve and guaranteeing a flat delivered number are commercially different animals, and separating the second lets an owner buy it from whoever will sell it. Wherever a guarantee sits, check that it travels. On an assignment, an OEM change of control, or a move to a new O&M contractor, a guarantee that does not survive the transfer was worth less than it looked on the day it was signed.

Where these packages fail

The expensive failure is a stack that does not line up. The offtake guarantees usable MWh at the POI on one schedule; the supply agreement guarantees retained capacity at the DC terminals on another; the two are measured at different boundaries, by different procedures, on different dates.

Whatever falls between them belongs to the project company — the one party in the stack with nobody to pass it to. The fix is unglamorous and it is engineering work, not legal work: one boundary, one test procedure, one set of reference conditions, written identically into every document that references the number.

The rest of the failures live in small print. A test written for a mild ambient and a fresh plant flatters the supplier and quietly resets the baseline the owner will be measured against for a decade. An exclusions clause phrased as a category rather than an enumeration — "events beyond the supplier's control" — can absorb almost anything, including a fleet-wide corrective campaign.

A sole-remedy clause with a low cap converts a guarantee into a small insurance policy, and once the cap is exhausted the shortfall is the owner's for the remaining term. None of this is exotic. It is the default drafting, and it changes only when someone asks.

Common misconception

A warranty and a performance guarantee are the same thing.

In reality: A workmanship warranty promises the equipment is free of defects and gets repaired when it is not. A performance guarantee promises measurable output — capacity, efficiency, availability — against defined tests, with money attached to the shortfall. A battery can be defect-free and still miss its guaranteed numbers; the two promises are written, tested and remedied separately.

Go deeper

Performance Guarantee, in context.

The Grid-Scale BESS course covers performance guarantee — and the rest of the system — from the ground up, the way it actually gets deployed.

Browse the course