A supercapacitor UPS does not wear out the way a battery does. Its capacity does not collapse, and its chemistry gives plenty of warning. What happens instead is a slow drift: internal resistance creeps up, capacitance creeps down, and the point at which the pack can no longer hold the load arrives gradually. The two things that set how fast that happens are the temperature the cells sit at and the voltage they are held at. Cycling, the factor everyone asks about first, matters far less.
That has a practical consequence for anyone running a supercapacitor UPS in an AI rack: the maintenance plan is not built around cycle counts, it is built around temperature, voltage and a periodic measurement of two parameters.
What “End of Life” Actually Means
There is no single number stamped on a supercapacitor datasheet that says how long it lasts, because lifetime depends on how it is used. What manufacturers do instead is publish a limited-life rating at maximum temperature and voltage, and define the end point by two measurable parameters.
The common industry definition is a loss of 20% of capacitance or a doubling of internal resistance, whichever comes first. In practice capacitance usually reaches that limit first. Maxwell’s own application data notes that capacitance typically falls to the −20% level before ESR has risen by 100%, which makes capacitance the parameter that normally triggers end of life.
How much margin you need depends on what the UPS is protecting. Eaton publishes different assumptions by application, and they are worth knowing because they set the design window rather than a hard limit:
| Application | Capacitance drop assumed | ESR rise assumed |
|---|---|---|
| UPS and medical, 10-year horizon | 20% | 1.7 times initial |
| Industrial regulators and power supplies | 30% | 2 times initial |
| Automotive | 50% | 3 times initial |
The pattern is deliberate. Where a missed backup is a safety event, the assumed degradation is small and the replacement is early. Where the load is tolerant, the pack is run further into its degraded state before anyone acts.
The Two Things That Age a Cell
Temperature and voltage dominate. Both act on the electrolyte, and both are governed by the same doubling behaviour.
EPCOS data put a 2.5 V cell held at 25 °C at a rated life of about ten years, with that figure doubling for every 10 °C reduction in operating temperature, or for every 100 mV reduction in applied voltage. The temperature part is the familiar rule borrowed from aluminum electrolytics, expressed as:
Lx = L0 × 2(T0 − Tx) / 10
So a cell rated 5,000 hours at 65 °C becomes roughly 20,000 hours if it actually runs at 45 °C. The same lever applies to voltage, which is why derating is the single most effective thing a designer can do. Holding a 2.7 V cell at 2.5 V rather than at its rated voltage roughly doubles the expected life on its own, and running it at 85% of rated voltage is a common recommendation for exactly this reason.
| Operating temperature | Relative life |
|---|---|
| 65 °C (rated) | 1x |
| 55 °C | 2x |
| 45 °C | 4x |
| 35 °C | 8x |
| 25 °C | 16x |
Inside a server rack, the cell temperature is not the same as the room temperature. A supercapacitor UPS module sitting next to a power supply, in a rack that already runs warm, can easily sit 15 to 20 °C above ambient. That difference is worth a factor of three or four in expected life, which is why mounting position inside the rack is a life decision rather than a mechanical one.
Why Cycling Is Not the Problem
Supercapacitors are specified for hundreds of thousands, often a million, charge and discharge cycles. That figure is real, and it is the reason they appeal for backup duty, but it describes cycle life, not service life.
A cell that is charged to its working voltage and simply held there still ages. This is calendar aging, and for a UPS application it is the dominant mechanism, because a backup pack spends almost all of its time sitting at voltage rather than cycling. A pack that is called on twice a year will reach its end of life at roughly the same point as one called on twice a day, if both sit at the same temperature and voltage.
There is one indirect link. Heavy ripple current through the pack raises internal heating, and internal heating raises cell temperature, so a poorly filtered duty can shorten life through the temperature term rather than through the cycling itself.
What to Measure, and How Often
Two parameters carry almost all the information. Capacitance indicates how much energy the pack can still deliver, and ESR indicates how much of that energy is lost before it reaches the load, especially at the moment of switchover when the current is highest.
Neither needs a lab. Capacitance can be estimated by discharging the pack into a known load while logging voltage against time, then taking C from the slope. ESR is measured by applying a current step and dividing the resulting voltage step by the current, which is what a purpose-built capacitance tester or a four-wire measurement does. A 1 kHz AC measurement gives a comparable number and is usually what a bench LCR meter reports.
- Measure on receipt, before the pack goes into service. Without a baseline, a later reading tells you nothing.
- Re-measure at a fixed interval, ideally at the same temperature and state of charge each time, so the numbers are comparable.
- Track the trend rather than the absolute value. A pack that has moved from 100% to 92% of nominal capacitance over three years is telling you something a single out-of-spec reading never will.
- Log temperature alongside. A pack that suddenly starts aging faster usually has a cooling problem, not a cell problem.
The interval depends on the application. For a rack UPS where a missed transfer is a real event, an annual measurement is a reasonable starting point, tightened to twice a year if the pack runs warm.

When to Replace
The textbook trigger is capacitance at 80% of nominal, or ESR at twice its initial value. In service, those numbers should be treated as the point at which the pack is formally at end of life rather than the point at which action becomes urgent, because the more useful question is whether the pack can still do its job.
A supercapacitor UPS is sized with margin, so a pack that has lost 15% of its capacitance may still hold the load for the full backup window with room to spare. Working that out means re-running the original sizing calculation with the measured values instead of the nominal ones, which is the same calculation covered in our guide to sizing a supercapacitor UPS. If the degraded pack still meets the window, the sensible action is to keep it and shorten the measurement interval, not to replace it on a date.
One thing to watch in a series string. Cells age at slightly different rates, and the weakest cell sets the string voltage window through the balancing circuit. Replacing a single cell in an aged string with a new one leaves the new cell carrying a disproportionate share, and the string usually ends up worse than before. For a rack UPS, replacing the whole module is normally the right call.
How This Differs from Battery Maintenance
A battery-backed UPS is maintained around failure. Capacity holds up and then falls away, internal resistance can rise sharply near the end, and the system relies on a battery management system to flag the transition. Maintenance is scheduled around an expected replacement date.
A supercapacitor pack behaves in a way that suits condition-based maintenance. The degradation is monotonic and measurable from an early stage, there is no equivalent of a sudden internal short, and the end point can be projected from a couple of readings rather than discovered. Maintenance is still needed, but its shape changes: fewer surprises, more trend lines.
For a rack that cannot tolerate an unplanned transfer, the practical programme is three items. Mount the pack where it runs cool. Hold it below rated voltage. Measure capacitance and ESR on a fixed schedule against a recorded baseline, and re-run the sizing calculation before deciding to replace anything.
FAQ
How long does a supercapacitor UPS last?
Around ten years is a reasonable expectation for cells held near 25 °C at a derated voltage, and longer if they run cooler. Real service life of ten to fifteen years at room temperature is quoted by manufacturers, with the figure halving for every 10 °C above the reference.
Do supercapacitors wear out from being charged and discharged?
Not in any practical sense. They are rated for hundreds of thousands to a million cycles. What ends their life in a UPS application is calendar aging at temperature and voltage, which continues whether the pack is cycled or not.
What is the end-of-life criterion for a supercapacitor?
A 20% loss of capacitance or a doubling of ESR, whichever comes first. Capacitance normally reaches the limit first. Some applications use tighter or looser assumptions depending on how much margin the load can tolerate.
Can I tell a supercapacitor is failing from its voltage?
No. A degraded cell still charges to the same voltage, it simply holds less energy and loses more of it in ESR. The change only shows up when you measure capacitance and ESR directly, or when the pack is asked to carry a load it can no longer support.
Xuansn supplies the supercapacitor cells and modules used in backup and buffer duty, across the supercapacitor series, with the rated life and temperature data you need to set a maintenance interval. If you are specifying a replacement pack or working out how much margin your design has, send the application, the backup window and the operating temperature to coco@xuanxcapacitors.com, or contact us here, and we will confirm a part and the life figures that go with it.





