BESS warranty forensics
The company that guarantees
your battery drew the curve.
Kyklios is the independent read of the capacity warranty you were offered: the conditions that quietly void it, whether your own declared dispatch violates them, and whether the guaranteed curve survives the physics of your site. Before you sign, not after the claim is denied.
Pre investment decisions. LFP chemistry. Independent of every OEM.
Illustrative study
Drag the dispatch. The guaranteed curve does not move.
| Year | 2 | 4 | 6 | 8 | 10 | 12 | 14 | 16 | 18 | 20 |
|---|---|---|---|---|---|---|---|---|---|---|
| Guaranteed | 90.5 | 86.6 | 83.6 | 81.0 | 78.8 | 76.8 | 74.9 | 73.2 | 71.6 | 70.0 |
| P90 | 92.8 | 89.0 | 85.8 | 82.8 | 80.1 | 77.5 | 75.0 | 72.6 | 70.3 | 68.0 |
| Margin P90 | +2.3 | +2.4 | +2.2 | +1.8 | +1.3 | +0.7 | +0.1 | -0.6 | -1.3 | -2.0 |
This is the outcome that matters and the one nobody prices in: the same battery, under the same guaranteed schedule, clears it or breaches it depending on how you dispatch it. The curve the OEM gave you assumed a profile. Yours is in your revenue model, and the two were never compared. That is why a verdict without a scenario is worth nothing, and why we will not hand you one.
Synthetic example, illustrative shapes only, not the calibrated engine. Cerro Verde BESS, 10 MW / 40 MWh, LFP 314 Ah, annual schedule to year 20. A real study models your cell, your site and your contract.
The gap
The OEM guarantees a curve. Nobody reads the contract.
A capacity warranty is a legal instrument with exclusions, measurement rules and a penalty ceiling. It is negotiated by people who are not battery physicists, against a curve produced by the party that also pays the claim. Three things go unchecked.
The conditions that void it
Temperature envelopes, C-rate ceilings, SoC dwell limits, cycle counts. The dispatch you declared in your revenue model can violate the warranty you were offered, and nobody runs that check before signing.
How SOH is actually measured
The contract defines the capacity test that fixes SOH at t=0 and at every checkpoint. Change the measurement protocol and the same battery passes or fails. This is where claims are won and lost.
The ceiling versus the loss
Liquidated damages are capped. The energy you do not have is not. When the penalty ceiling sits below the real revenue loss, the warranty transfers less risk than the model assumes.
The 15 to 20 year commitment
The market is writing degradation into the contract.
A revenue model spreading across every continent in storage is a fixed payment for capacity that is there when called: capacity markets, availability tenders, tolling, CfDs. The payment is flat. The risks that erode it are not, and every one of them transfers to you for fifteen to twenty years. An OEM warranty is one battery against one curve. This is the same commitment written larger, and it rests on three things almost nobody models before bidding.
Degradation
The capacity you sold shrinks from the day it is energised, at a rate set by your chemistry, your temperatures and the dispatch you chose. The contract does not ask why. It settles against the MW you declared in year one, every year, to the end of the term.
Augmentation
Committed capacity plus a shrinking system means augmentation is not a remedy for a bad year. It is a planned obligation: when you add MWh, how many, at what cost, and whether oversizing on day one beats augmenting in year six. That trade off is calendar ageing against cycling ageing, which is precisely the arithmetic a degradation model exists to do.
Availability
These contracts penalise unavailability against an index that you declare when you bid. The most detailed public precedent sets the floor at 95 percent with damages at twice the capacity charge. The number you write down becomes your penalty benchmark for the life of the contract, and most bidders inherit it from a spreadsheet nobody can trace.
The same instrument, on every continent
None of this is our forecast. The instruments are public, current and converging.
United Kingdom
The Capacity Market clarified Permitted Augmentation for battery storage in its rules and re verifies capacity with an Extended Performance Test. Fail the test and the payments stop.
Italy
The transmission operator auctions storage capacity on fifteen year contracts, gigawatt hours at a time, with availability obligations attached.
India
Standard battery purchase agreements set an annual capacity test to an international standard, a round trip efficiency floor measured AC to AC at the meter, and a 95 percent availability floor with liquidated damages at twice the charge.
The commercial layer
Integrator warranties now run to fifteen and twenty year terms, and independent certifiers are hired to verify fleet availability claims. The market is asking someone to stand behind exactly this commitment.
Two different clocks
Efficiency is not capacity.
The efficiency threshold in these auctions is round trip efficiency, not capacity retention, and confusing them is the most expensive mix up in a bid. Efficiency is the transformer: how much each round trip loses, set mostly by the equipment chain, moving slowly. Capacity is the tank: how much the system can hold, shrinking every year with calendar and cycling. Two numbers, two clocks, two different physical causes. A bid that treats them as one is wrong on day one, in whichever direction turns out to be expensive.
One limit, stated before you ask: our engine models degradation. It does not produce an availability forecast, and we will not dress one up as engineering. What a study contributes on availability is contractual arithmetic: what your declared index costs you in the years you miss it, and where the contract hides that formula.
The auction questions, added to the sevenWhy not just
The alternatives, and where each one leaves you.
Kyklios is not competing with your independent engineer and not competing with the OEM's laboratory. It sits in a gap that the other four options leave open.
- Trust the OEM's curve?
- It was produced by the party that pays the claim, for a generic duty cycle, not yours. Nobody in the chain has tested it against how you actually plan to dispatch. The one who guarantees cannot audit himself.
- Wait for the lender's engineer?
- They arrive at due diligence, once the supply agreement is signed. By then the schedule, the exclusions and the penalty cap are fixed. The leverage was spent months earlier.
- Build it in house?
- Two hires with cell-level expertise plus a calibration programme, and roughly eighteen months before the first usable answer. For a handful of decisions a year.
- Do nothing?
- This is the real alternative most projects pick, usually without deciding to. You transfer the degradation risk by contract, and that transfer is worth exactly what the clauses nobody read say it is worth.
The deliverable
A study you can hand to a lender's engineer without flinching.
Two to three weeks from a complete intake. Fixed scope, fixed price, 30 to 50 percent up front.
Eligibility check
Your declared operating profile against the conditions that void the warranty. The cheapest finding in the study and usually the one that moves the negotiation.
Degradation model
Calendar and cycling fade for your cell, your site temperature and your duty cycle, rolled up from cell to usable AC energy. P50 and P90 trajectories to year 20.
Conditional verdict
Pass or at risk at every contractual checkpoint, evaluated at P50 and at P90, with the margin stated in points of SOH. Conditional on an operating scenario you sign off.
Warranty gap report
Where the modelled curve crosses the guaranteed schedule, which checkpoints are thin, and what the penalty formula actually pays at each breach.
Augmentation calendar
Not a remedy, a schedule: when the system drops below its service requirement, how many MWh to add, whether oversizing now beats augmenting later, and nominal CAPEX against a dated, hashed price index. Costs are nominal by design. We do not sell an augmentation NPV.
Audit manifest
Model version, parameter hashes, and the source of every single input with its URL, DOI or invoice. Verifiable with sha256 and the manifest alone. No access to us required.
Why the bands are visible
We show you the uncertainty because the uncertainty is the product.
Anyone can hand you one line and call it a forecast. A twenty year extrapolation of cell physics has a real band around it, and a study that hides that band is selling comfort, not engineering. Kyklios states the width, states where the model stops being trustworthy, and refuses to call something a verdict when the margin is narrower than the band. The limits are printed on the cover of every report, not buried in an appendix.
Scope and limitsDeterministic
Same input, byte identical output. No black box, no model that quietly changes under you.
Provenance or it does not exist
Every coefficient carries a citation. No parameter compiles without a traceable public source.
Independent by construction
We audit the contract and the coherence of the OEM's own curve. We do not compete with the OEM's laboratory, and we do not represent any OEM.
Two ways to work with us
Developers and IPPs
You have an OEM offer on the table and a decision to make before due diligence. The study gives you the questions to ask, the leverage to renegotiate the schedule, and a defensible position in your own investment committee.
Start a conversationIndependent engineers
You have the client, the signature and the insurance, but no battery physics practice. We are the engine behind your report. You keep the relationship and the seal.
See how the channel worksNext step
A short call, and an honest answer about fit.
Thirty minutes. Bring the OEM offer and your duty cycle assumptions. If the study will not move money for you, we will tell you on that call rather than sell you one.
Start a conversation