Service Robot Battery Warranty: Cycle Life and Claim Evidence
At a glance: A robot battery warranty sounds like a single promise about capacity. In practice it is three separate promises, about capacity, about cycle count and about usable shift length, and the three rarely expire together. This guide unpacks how each is defined, why the definition of a “cycle” decides most disputed claims, what evidence has to be collected before a claim exists, and the clauses worth writing into the purchase contract while there is still leverage.
What a Battery Warranty Actually Promises
Most pack warranties in a service robot purchase order run to a short paragraph, and that paragraph is usually doing three jobs at once without saying so. Reading it as one number is the first mistake buyers make.
The first promise is capacity retention. The pack will still deliver some stated proportion of its original usable energy, measured under defined conditions, at the end of a defined period. This is the promise people picture, and it is the one least likely to be enforced as written, because proving it requires a controlled discharge test that neither party wants to run.
The second promise is cycle life. The pack will survive a stated number of cycles within that period. This is the promise vendors headline, usually as a single figure such as 2,000 or 3,000 cycles, and it is the number that survives into the marketing page while the conditions attached to it do not.
The third promise is usable shift, which is usually implied rather than written. The robot will complete its expected duty cycle without an unplanned mid-shift charge for the term of the agreement. No datasheet states it in these words, but it is the promise the operation actually depends on, and it fails first.
The three diverge because they are measured on different clocks. Capacity is measured on energy, cycles are counted on throughput, and shift coverage is measured on time and duty mix. A fleet can be comfortably inside its capacity warranty and its cycle warranty at the same moment it has stopped completing shifts, which is precisely the situation where a buyer discovers the warranty does not address the problem they have.
An operator in year three of a cleaning contract typically does not care about a percentage on a test bench. They care that a machine which used to finish a floor in one charge now needs a mid-afternoon top-up, which converts a one-shift unit into a two-charge unit and quietly removes capacity from the plan. That is a shift failure, not a capacity failure, and it is worth knowing which instrument in the contract responds before the argument starts.
Defining a Cycle When the Robot Never Fully Discharges
The single most consequential definition in the warranty is the word cycle, and it is almost never defined. For an electric vehicle the ambiguity is tolerable, because a full discharge is a recognisable event. For an opportunity-charged robot fleet it is not: units return to a dock between tasks, top up from 60 to 90 percent, and leave again.
The convention that most battery management systems actually count is the equivalent full cycle, sometimes written EFC. One equivalent full cycle is the cumulative energy throughput equal to one full pack capacity, regardless of how many partial excursions it took to get there. Two top-ups each delivering 30 percent of pack capacity are 0.6 equivalent full cycles, not two cycles.
This matters because the two counting methods produce wildly different numbers, and a buyer reading a BMS counter without knowing which method it uses will draw the wrong conclusion about remaining life.
| Metric | What it counts | Typical robot value | Claim risk |
|---|---|---|---|
| Full-cycle count | Excursions from full to empty | Low; many fleets never reach it | Understates real wear, favours vendor |
| Equivalent full cycle (EFC) | Cumulative energy throughput ÷ pack capacity | The meaningful wear figure | The number to contract on |
| Cumulative Ah throughput | Raw charge and discharge amp-hours | Auditable from BMS logs | Best evidence for a claim |
| Charge events | Number of dock connections | High; unrelated to wear | Misleading if used as a proxy |
Work the arithmetic once and the problem becomes visible. Take a compact scrubber carrying a 1.2 kWh pack, doing three passes a day at 40 percent depth of discharge each. Each pass consumes roughly 0.48 kWh, so the daily throughput is about 1.44 kWh, which is 1.2 equivalent full cycles per day. Over 300 operating days that is 360 equivalent full cycles a year. If the pack is rated at 2,000 equivalent full cycles to the warranty threshold, the theoretical life is roughly five and a half years of operation, not the eight or ten years a buyer might infer from a five-year calendar warranty.
Run the same arithmetic on a heavier duty cycle and the conclusion changes materially. A large-format scrubber with a 3.6 kWh pack running two shifts at 70 percent depth of discharge consumes about 5.0 kWh a day, or roughly 1.4 equivalent full cycles, and at 320 operating days that reaches 448 equivalent full cycles a year, putting the same 2,000-cycle threshold at about four and a half years. The pack that looks identical on a datasheet has a materially shorter financial life because the duty cycle is different.
This is the reason a cycle warranty quoted without a stated depth of discharge is not really a warranty at all. Depth of discharge, ambient temperature and charge rate all move the achievable cycle count, and a vendor quoting 2,000 cycles is implicitly quoting a test condition. The reference conditions in most cell datasheets are a full charge and discharge at a moderate rate at 25 degrees Celsius, which is a laboratory life rather than a site life.
Buyers should ask for the cycle figure to be stated together with the depth of discharge, the temperature band and the charge rate at which it applies, and should ask which counter in the BMS the warranty will be measured against. If the answer is vague, the clause will be read in the vendor’s favour at claim time, because they will be holding the logs.
Degradation Thresholds: 70 Percent or 80 Percent, and What It Does to Coverage
The threshold at which a pack is declared end of life is a commercial decision dressed as a technical one, and the choice between 70 and 80 percent of original capacity has a direct operational consequence that is rarely spelled out.
Eighty percent is the common automotive convention and it is increasingly used for robot packs, because below it the internal resistance rise becomes visible in charge time and in voltage sag under load. Seventy percent is the older industrial convention, more permissive and cheaper for the vendor to honour, and it typically buys another one to two years of calendar life before the pack is formally out of warranty.
The operational translation is what matters. A pack at 80 percent capacity does not deliver 80 percent of the original shift. It delivers less, because the robot also has to carry reserve, operates at lower average voltage and spends longer charging. In practice the usable duty reduction runs ahead of the capacity percentage, so a pack at 80 percent of rated capacity commonly delivers something closer to 70 to 75 percent of its original shift coverage once charge time is netted off.
| Warranty threshold | Typical year reached (EFC basis) | Shift impact | Who it favours |
|---|---|---|---|
| 90% | Year 1–2 | Negligible | Buyer, if it can be enforced |
| 80% | Year 3–4 | Mid-shift top-up begins | Balanced; the useful line |
| 70% | Year 4–6 | Shift no longer completed on one charge | Vendor; defers replacement cost |
The practical recommendation is to negotiate the threshold at 80 percent and to state explicitly that it is measured as capacity at a defined rate and temperature, not as an estimate from voltage. A threshold without a measurement method is a number both parties will read differently, and the party holding the data reads it last.
There is a second threshold worth defining alongside it: the capacity floor for the duty. This is the point at which the pack can no longer complete the contracted duty cycle, and it is normally reached after the warranty threshold has already expired. Writing it into the contract as a separate, later trigger is what converts the warranty from an asset guarantee into an operational one, and it is the clause that actually protects the service obligation. The interaction between these two dates is the same arithmetic that drives the replacement planning horizon described in the guide to battery health and replacement planning.
Cycle-Count Warranty or Throughput Warranty: Which Instrument Protects You
A cycle-count warranty and a throughput warranty protect different parties, and confusing them is common. The distinction is worth understanding before the commercial negotiation, because the two instruments fail in opposite directions.
A cycle-count warranty is written in the language of the pack: 2,000 equivalent full cycles, measured on the BMS counter. It is easy to audit because the counter is unambiguous once the counting method is fixed, but it is disconnected from the operation. A fleet that sits idle for six months ages anyway, and a fleet that runs short light passes accumulates cycles slowly while its calendar life drains. The warranty quietly runs out on the calendar without the cycle limit being approached.
A throughput warranty is written in the language of the operation: the pack is warranted to deliver a stated cleaned area, moved pallet volume or operating hours before the capacity threshold is met. This is the instrument a buyer should prefer, because the failure it addresses is the one the business actually experiences, and because it makes the vendor’s exposure proportional to the work performed rather than to a counter.
Vendors tend to prefer cycle counts because the counter is under their instrumentation. Buyers should prefer throughput because the service obligation is what they are paying for. The pragmatic outcome is usually a warranty carrying both limits, with the buyer arguing for the throughput limit to be the binding one and the calendar limit to be generous enough not to bind first.
| Dimension | Cycle-count warranty | Throughput warranty |
|---|---|---|
| Unit of measure | Equivalent full cycles on BMS | Area, volume or hours delivered |
| Ease of audit | High, once counting method is fixed | Medium, needs operational records |
| Aligned to business risk | Weakly | Directly |
| Effect of idle time | Warranty expires unused | Warranty expires unused |
| Effect of heavy duty | Threshold reached early | Threshold reached early |
| Preferred by | Vendor | Buyer |
Neither instrument covers calendar ageing well, and calendar ageing is real: lithium packs lose capacity sitting at high state of charge in warm rooms whether or not they are cycled. A fleet in a hot plant room, charged to full overnight and rarely used, can lose capacity on the calendar faster than a busy fleet in a cool building. The purchase contract should therefore carry a calendar limit stated in months with a defined storage state of charge, not simply a term of years, and it should say who is responsible for storage conditions when the site is at fault.
The Claim Evidence File You Have to Build From Day One
Battery claims are won on records, and the records have to exist before the failure. This is the part of the process that buyers most often skip, because in year one the pack works and the logs seem like housekeeping. By year three, when the claim is live, the ability to reconstruct the counter history is the difference between a replacement and an argument.
What a defensible file contains is not complicated, but it has to be collected on a cadence rather than at the moment of dispute.
- BMS counter exports at a fixed interval. Monthly is adequate for most fleets and weekly for heavily used ones. The export must include the equivalent full cycle counter, state of health, and cumulative charge and discharge throughput. Screenshots of an app are not an export.
- State of health trend, not a single reading. A claim needs the slope, not the endpoint, because the slope distinguishes a pack that failed from a pack that aged normally.
- Cell balance and maximum delta. A pack whose cells drift apart fails before the aggregate capacity suggests; the maximum cell voltage delta is the early evidence that a string is degrading faster than the pack average.
- Charge and discharge temperature records. Ambient and pack temperature history is the vendor’s first line of defence, because high-temperature operation genuinely does accelerate wear. A buyer who cannot show the pack ran within its stated band hands over a legitimate exclusion.
- Depth of discharge distribution. Not just the total cycles but the shape of the excursions, which supports the equivalent-full-cycle arithmetic and rebuts a suggestion of abusive deep cycling.
- The operating duty log. Hours, area or volume delivered per day, from the fleet software rather than from a manual sheet, so the throughput warranty can be computed independently.
- The commissioning record. The baseline capacity measured at handover, which is the denominator every later percentage is relative to. Without it, 80 percent of what is an open question.
The commissioning item deserves emphasis because it is cheap at the start and impossible to reconstruct later. A controlled discharge capacity test at handover, recorded with the method the vendor accepts, establishes the baseline and also establishes that the buyer knew how to test. The same acceptance gate that produces this record is described in the guide to acceptance testing and commissioning, and it is worth tying the two together so the test is not run twice.
There is a contract point hiding in this list. The BMS data belongs to whoever can export it, and in a RaaS or managed-service structure that is often the provider. A buyer who does not have standing access to monthly exports cannot build any of the above. The right of access, in a machine-readable format, at a defined interval, and surviving termination, belongs in the contract alongside the warranty itself, and the wider question of who owns fleet data is set out in the guide to data ownership and API interoperability.
Pro-Rating, Exclusions and How a Claim Actually Pays Out
The gap between a warranty clause and a cheque is filled by the pro-rating formula, and the formula is usually more favourable to the vendor than the headline term suggests.
Most battery warranties are written on a declining schedule. A five-year warranty might pay 100 percent of replacement cost in year one, 60 percent in year two and 20 percent by year five, which means a failure at the point where the pack is most likely to fail is the point at which the warranty pays least. This is not unreasonable, since the buyer has had use of the asset, but it should be modelled rather than assumed, because the residual cost of an unplanned pack replacement in year four is often larger than the buyer expected.
| Claim element | Typical contract position | Where the value erodes |
|---|---|---|
| Parts cost | Covered on a declining percentage | Late-term failures pay a small fraction |
| Labour cost | Often excluded or capped | Site labour and downtime are rarely recoverable |
| Freight and handling | Frequently excluded | Battery transport is regulated and expensive |
| Downtime | Almost never covered | The real loss is the missed service, not the pack |
| Diagnosis time | Sometimes chargeable if no fault found | Disputed claims can cost the buyer money |
Freight is worth a specific mention because it is disproportionately expensive for this component class. Lithium packs are regulated for transport, which means packaging, labelling and carrier restrictions that ordinary spare parts do not face, and a warranty that covers the part but not the regulated transport can leave the buyer with a meaningful bill. The transport requirements are summarised in the guide to battery certification and transport.
The exclusions are more predictable than buyers expect. High-temperature operation, water ingress beyond the stated rating, charging with non-approved equipment, deep discharge from prolonged storage, physical damage and unauthorised repair cover the overwhelming majority of declined claims. Each is defensible in principle, and each becomes a dispute only when the buyer has no records to contest it, which returns the argument to the evidence file.
One exclusion deserves specific attention in the contract: the definition of normal use. If the vendor defines normal use as the duty cycle of a standard deployment and the buyer runs two shifts in a warm environment, the warranty may be void from the first week in ways neither party intended. Duty cycle assumptions should be written into the schedule that forms part of the contract, with the buyer’s actual pattern recorded, so the warranty is calibrated to the deployment it will actually see.
What to Write Into the Purchase Contract
Six clauses carry most of the value, and they are far easier to negotiate before signature than after a failure.
A defined cycle. State that the warranty is measured in equivalent full cycles as computed from cumulative energy throughput divided by rated pack capacity, and name the BMS field that will be used. This single clause removes the most common source of dispute.
A stated threshold with a stated method. Capacity measured at a defined rate and temperature, at 80 percent, with the measurement procedure attached as a schedule rather than referenced loosely.
A throughput limit alongside the cycle limit. Area, volume or operating hours, with the buyer arguing that the throughput limit binds first.
A calendar limit with storage conditions. Months, not just years, with a defined storage state of charge and a division of responsibility when the site controls the environment.
Data access as a warranty condition. Monthly machine-readable BMS exports at no cost, available to the buyer directly and surviving termination of a service agreement, because a warranty that cannot be evidenced is not a warranty.
A defined claim and response timeline. Hours to acknowledgement, days to a replacement decision, and a remedy if the vendor misses it, which is usually a credit rather than a further promise.
The commercial logic is that the warranty is worth roughly the replacement cost of one pack multiplied by the probability that the vendor accepts the claim. Improving the second term through definition is usually cheaper than improving the first through negotiation, and the definition work costs nothing but attention at contract stage. The way these clauses interact with the wider service agreement, including the boundary between warranty and a paid service contract, is set out in the guide to service contract versus warranty economics, and the mechanics of lodging and tracking the claim itself are covered in the guide to warranty claims and the RMA process.
Where the Battery Warranty Sits in the Fleet Cost Model
A battery replacement is a predictable capital event and it belongs in the ownership model from the first year, not in the contingency line. The arithmetic above gives a rough date: annual equivalent full cycles, derived from duty cycle and pack size, divided into the warranty threshold, converted into years by operating days.
For a compact indoor cleaner at roughly 360 equivalent full cycles a year against a 2,000-cycle threshold, the pack is a year five to six event. For a large-format scrubber on two shifts, it is a year four event. For a lightly used delivery unit, the calendar limit binds before the cycle limit and the pack is a year five event regardless of use. None of these are emergencies if they are in the model.
The second cost to include is the interim degradation cost, which is the coverage lost between the point where shift length starts to fall and the point where a replacement is justified. Running a pack from 90 to 80 percent state of health may cost a mid-shift top-up or an extra unit to cover the gap, and that cost is incurred for a year or more before any replacement decision is due. It is easy to miss because it shows up as labour and scheduling friction rather than as a line item.
Both figures belong in the same ledger as maintenance, energy and insurance. The full structure is in the total cost of ownership breakdown, and the pack-side specification decisions that determine which of these dates you land on are in the guide to battery pack specification and sourcing and the duty-cycle analysis in battery chemistry by duty cycle.
If you are working through a battery warranty on a specific fleet, we can provide the BMS field list, a commissioning capacity test procedure and the warranty schedule wording we use, so the baseline exists before the first claim does. The cleaning robot and delivery robot ranges both carry the same pack-level documentation, and the quote form reaches a technical contact directly.
