Skip to content

Free. No email required.

The seven questions to ask before you sign a capacity warranty.

These are the questions a battery warranty rarely survives. Ask them of any manufacturer, in any market, with us or without us. If the answers are good, you have a good contract. If they are not, you have just found your negotiation.

We deliberately name no manufacturer. This is not an accusation: a capacity warranty is a normal commercial instrument and the OEM is doing its job, which is to bound its own liability. The problem is that nobody on your side of the table is asking.

  1. Which operating conditions void this warranty, and does my own declared dispatch violate any of them?

    Why it matters

    The guarantee comes with an operating envelope: temperature limits, C-rate ceilings, state-of-charge dwell limits, cycles per year. Your revenue model was built by someone else, usually with a more aggressive dispatch. The two documents are almost never read side by side.

    What a good answer looks like

    A table of conditions with numeric limits, and a written confirmation that your declared profile sits inside all of them.

  2. How exactly is state of health measured, at t=0 and at every checkpoint?

    Why it matters

    The contract defines the capacity test that fixes SOH. Rate, temperature, rest time, voltage window, whose instruments, who witnesses it. Change any one of them and the same battery passes or fails.

    What a good answer looks like

    A named test procedure, referenced as an exhibit, with ambient conditions and tolerance stated. Not "as per manufacturer standard practice".

  3. What shape is actually guaranteed: an end-of-life threshold, an annual schedule, or a penalty formula?

    Why it matters

    These three transfer completely different amounts of risk. A single threshold at year 20 leaves you unprotected for nineteen years. An annual schedule is enforceable every year.

    What a good answer looks like

    An annual schedule with a minimum SOH per year and a defined remedy at each checkpoint.

  4. What does the penalty actually pay, and where is it capped?

    Why it matters

    Liquidated damages are capped. The revenue you lose is not. If the cap sits below your real exposure, the warranty transfers less risk than your model assumes, and the difference is yours to carry.

    What a good answer looks like

    The formula, worked through on your own numbers, at each checkpoint, against the cap. If nobody has done that arithmetic, it has not been negotiated.

  5. If the system falls below the curve, who pays for augmentation, and on what price basis?

    Why it matters

    A remedy is not the same as compensation. Adding MWh in year 12 at year 12 prices is a very different contract from a cash payment against a cap fixed today.

    What a good answer looks like

    A named remedy, a price basis, and a stated timeline for execution.

  6. Who is standing behind this in year 12, and with what instrument?

    Why it matters

    A twenty year guarantee is only as good as the balance sheet behind it. Cell manufacturing consolidates. The entity that signed may not be the entity that answers.

    What a good answer looks like

    A parent company guarantee, a warranty bond or a standby letter of credit, with a defined duration. A signature is not an instrument.

  7. What can suspend or void it that is not my fault?

    Why it matters

    Curtailment, grid events, the EMS you chose, a firmware update, the availability of spares, a capacity test delayed by the supplier. Exclusions written for the supplier's convenience become your risk quietly.

    What a good answer looks like

    A closed list of exclusions, not an open one. "Including but not limited to" is where a guarantee goes to die.

Added for capacity auctions and tolling

Three more, if your revenue is an availability payment.

If you are bidding a capacity auction or signing a long term availability or tolling agreement, the seven questions above still apply. The difference is that there may be no OEM across the table: the counterparty is your own bid. These three are addressed to it.

  1. Does my declared efficiency trajectory clear the threshold in every contract year, measured where the contract meters it?

    Why it matters

    These contracts settle round trip efficiency AC to AC at the meter, year after year. A brochure number is a DC number under laboratory conditions. Declaring it optimistically is not a rounding error, it is a recurring invoice: the gap between declared and delivered comes back every settlement period for the life of the term.

    What a good answer looks like

    A trajectory built from the full efficiency chain at the point of interconnection, with the margin to the threshold stated for every contract year. Not one DC figure copied twenty times.

  2. What does my self declared availability index cost me in the years I miss it?

    Why it matters

    Unavailability is penalised against the index you wrote into your own bid, and public precedent puts the damages at a multiple of the capacity charge. It is the one number in the deal you chose freely, it binds you for the whole term, and in most models nobody can say where it came from.

    What a good answer looks like

    The penalty formula worked through at your index minus one, minus two and minus five points, in your own numbers. If that table does not exist, the index was not chosen. It was guessed.

  3. Does my augmentation plan have dates and CAPEX, or is it an assumption?

    Why it matters

    Committed capacity plus a shrinking system equals mandatory augmentation. If the plan is a sentence in the base case rather than a schedule, the commitment is funded by hope, and no auction pays for hope.

    What a good answer looks like

    A schedule: the SOH trigger, the year, the MWh added, space and interconnection reserved, nominal CAPEX against a dated price index, and the oversize now versus augment later comparison actually run.

If the answers are uncomfortable, that is the finding.

Every one of these is answerable from the offer already on your table, and you do not need us to ask them. What a study adds is the arithmetic: your declared dispatch checked against the envelope, the guaranteed schedule tested against modelled degradation at P50 and P90, and the penalty formula worked through in your own numbers.

Have these answered for your offer