The Capacity Fade Warranty Clause AMR Buyers Need Before Fleet Deployment

The Capacity Fade Warranty Clause AMR Buyers Need Before Fleet Deployment

A capacity fade warranty is only as good as the test conditions stapled underneath it. Before a fleet of AMRs or AGVs leaves the dock, the clause has to lock down four verification parameters — charge C-rate, discharge C-rate, ambient temperature, and depth of discharge — along with the end-of-discharge voltage used for the capacity check, the number of full equivalent cycles at which the threshold applies, and whether retained capacity is measured at the pack terminals or summed at the cell level. A headline cycle-life number without those conditions attached is a marketing figure, not a remedy you can collect on.

Start with the duty cycle, not the cycle count

I have been in that room enough times to know where the language sits. Page fourteen, usually, in a paragraph that reads like boilerplate, wedged between shipping damage and spare parts pricing. Someone on your side has already found the headline: 3,000 cycles to 80 percent of beginning-of-life capacity. Heads nod. The procurement lead types it into the comparison sheet next to two other suppliers claiming something close to the same number.

Then I ask when, during a shift, the pack ever sees 1C.

That is the entire problem with cycle life taken at face value. The number was almost certainly established on a full-depth discharge, at a fixed C-rate, at a fixed temperature, on a fresh pack in a chamber. Your fleet does something else. Say your AMRs leave the dock at 90 percent state of charge, run a ten-hour shift with intermittent traction and lift loads that average well under 1C and peak above it for a few seconds at a time, come back at 25 to 35 percent, and dock again. That is not a full cycle. It is a partial cycle in the middle of the voltage window, repeated four or five times a day, which generally ages a pack differently than full-depth cycling — often more gently per unit of energy throughput, but with a growing calendar component, because in a two-shift operation the pack never really rests.

So when a supplier says 3,000 cycles, you cannot tell whether that means four years or seven in your building. That translation is your job. It happens before signature, not after the first pack comes up short and someone asks you why the warranty does not cover it.

What I want to see is a short appendix to the warranty exhibit. Call it Application Duty Cycle. It states, in writing, what the fleet actually does: nominal charge rate around 0.3C from 25 to 90 percent SoC; average discharge 0.4C with peaks to 1.2C for up to 30 seconds; ambient 15 to 32 degrees C, with excursions near a weld cell; a typical depth of discharge around 60 percent; opportunity charging at the dock, twice per shift; and the end-of-discharge voltage you consider empty.

Two things happen when you put that appendix in front of a supplier. Either they adjust the warranted cycle count downward because the real duty cycle is harder on the pack than the datasheet condition, or they keep the number and you have just made it enforceable. Both outcomes are better than the number you walked in with.

Write the reference capacity check into the clause

A retained-capacity threshold is meaningless without a method so specific that two different labs would get the same answer. Vague language like "capacity shall be verified by an agreed test method" is where claims go to die, because it defers the argument to the moment you are already in a dispute.

I want the clause to name:

• The charge rate and the CV cutoff current at which charging terminates.
• The rest period between charge and discharge.
• The discharge rate used for the check, which is not the same as your fleet's average discharge rate and should be stated separately.
• The end-of-discharge voltage, per cell or per pack, that defines the bottom of the check.
• The ambient temperature and its tolerance — plus or minus two or three degrees C is typical — held for the duration.
• The number of consecutive reference checks and which one counts. Best of three and mean of three are different clauses, and the gap between them can be a point or two of capacity.
• Who measures, with what instrument, and who calibrates it. External cycler and shunt, or the pack's own BMS, or both with the external instrument governing.

That last item pairs with a question most buyers never ask: is capacity measured at the pack terminals or summed at the cell level?

Insist on pack terminals, under load, measured at the output. Cell-level summation flatters the result. Interconnect resistance, contactor and fuse losses, and current-sense error in the BMS all shave real capacity off what the cells can deliver, and a capacity fade claim often lives or dies inside a two or three point margin. A supplier who tests at the cell level is not lying, but they are measuring a different product than the one bolted into your robot.

The threshold, the clock, and the proration schedule

The warranted metric is almost always retained capacity as a percentage of beginning-of-life capacity after a stated number of full equivalent cycles. Define the full equivalent cycle in the clause itself, as cumulative ampere-hours of throughput divided by rated capacity. Without that definition, you are at the mercy of whatever counter happens to be running — more on that below.

And define beginning of life as a measured value, not a nameplate. Two packs from the same build can leave the line a point or two apart, and storage or shipping can take another fraction before commissioning. Write it as the mean of three reference checks at acceptance, or the capacity verified during factory acceptance, recorded and countersigned by both sides. Keep that record somewhere you will still be able to find it in year three, because it is the number every capacity argument eventually circles back to.

Add a calendar line running in parallel with the cycle line — capacity below 80 percent at 2,000 full equivalent cycles or 36 months from commissioning, whichever comes first. Be aware that "whichever comes first" cuts both ways. If your fleet runs hard, the cycle line will dominate and the calendar line is decoration. If your fleet sits idle in a seasonal operation, the calendar line is the only thing protecting you. Know which fleet you have before you negotiate.

Then there is the remedy, and this is where capacity fade warranties quietly turn into discounts. A proration schedule — something like 100 percent credit toward a replacement pack in year one, 60 percent in year two, 35 percent in year three — converts a performance failure into a partial coupon. That credit is usually applied against the supplier's list price for a replacement, and it rarely covers freight, the labor to swap packs across forty robots, or the throughput you lose while the fleet is down.

So push in two directions. First, get freight and a defined response and replacement time written in, because on a fleet the downtime is the expensive part, not the hardware. Second, carve early failures out of the proration entirely. A pack that drops below a hard floor inside the first 300 full equivalent cycles, or inside the first six months, should trigger full replacement. That is not a performance warranty anymore; that is a defect, and it deserves different treatment than normal wear.

Exclusions that kill a claim before you file it

Read the exclusion list as a list of things you must be able to prove you did not do.

Storage and transit temperature is the one that catches people. The clause says claims are void if the pack is stored above a stated temperature or below a stated state of charge, or if it is charged outside an ambient window, or if it is charged with equipment the supplier did not approve. Then your packs sit in a trailer in a Phoenix yard for three weeks in July, or in a warehouse with no HVAC, and nobody recorded anything. You can lose a legitimate capacity fade claim on evidence you never collected.

That cuts two ways. Get the storage and transit limits written down in plain numbers, and then write the measurement obligation into your own receiving process — an inbound temperature log, a state-of-charge check, a photo of the label. If it is not logged, assume the exclusion applies to you.

The field data clause: who pulls the logs, and what counts as a cycle

Many warranties now include a clause letting BMS-recorded cycle counts and energy throughput support or deny a capacity fade claim. On its face that sounds neutral and even convenient. It is not neutral, and it is only convenient if you control the export.

I have seen a BMS whose cycle counter incremented on every return to dock. The fleet had accumulated roughly 900 full equivalent cycles by throughput, and the log read a shade under 2,400. If your warranty measures the claim against that counter, you burn the entire cycle allowance in year two and you will lose the argument, because the clause says the supplier's records govern.

So write it this way. The BMS must record cumulative charge and discharge ampere-hours, not merely an event counter. You get the raw log format and a means to read it that does not depend on a supplier-licensed tool being available to you forever, or a license that survives the end of the agreement. Raw logs are the evidence; the supplier's records are a cross-check. And archive the export monthly, on your side, somewhere that is not one engineer's laptop.

What I do before the signature

I take the pilot fleet's logs, convert them into the test conditions written in the clause, and see whether the two describe the same product. If the fleet runs 60-point partial cycles in a 20 to 32 degree warehouse and the clause verifies at 1C full depth at 25 degrees in a chamber, the clause is describing a different life than the one you are buying. Redline it, or accept that the number is aspirational.

And whatever the final language says, I keep the acceptance capacity checks and the monthly exports. The first thing that gets argued about when a pack comes up short is not the threshold. It is what the pack started at.