UPS Battery Aging: Why Runtime Minutes Drop Over Time
Introduction #
UPS battery aging is why a string that delivered 30 minutes at commissioning can fail a 20-minute acceptance test three years later—even when float voltage still looks “green.” Capacity fades from calendar life, cycle wear, heat, and rising internal resistance. Operators often blame the UPS inverter first; the DC bus usually fails quietly first.
When this guide fits: You need to explain runtime degradation to operations or finance, set EOL Ah margin, decide when to refresh strings, and keep contract minutes honest across the string’s life.
When it does not fit: Day-one Ah sizing from kW and minutes alone—start with How to calculate UPS battery size. Field test schedules and CMMS templates belong in UPS Battery Maintenance. Temperature-only derates: UPS battery temperature derating.
Pair every aging assumption with the UPS Runtime Calculator and UPS Battery Calculator. Theme hub: UPS calculator.
Why aged batteries lose minutes #
Usable energy falls for four overlapping reasons. Treat them as budgets, not mutually exclusive checkboxes:
| Mechanism | What changes | Runtime effect | Early signal |
|---|---|---|---|
| Calendar fade | Active material / dry-out on float | Fewer Wh at same nameplate Ah | Slow Ah drop on load tests |
| Cycle wear | Deep or frequent discharges | Faster fade on transfer-heavy sites | Event count + DoD history |
| Impedance rise | Higher internal resistance | Voltage hits cutoff earlier under load | Ohmic / impedance trend |
| Temperature history | Hot float accelerates all of the above | Same Ah math, shorter minutes | Cabinet max temp logs |
Rule: Treat nameplate Ah as day-one optimistic. Design and acceptance should use EOL Ah (often 70–80% of day-one, or the OEM table). Finance plans refresh when impedance or load-test minutes cross policy—not when the first outage fails.
Calendar vs cycle: two clocks #
A VRLA string in a quiet IT closet may die on calendar (years on float at elevated temperature). The same chemistry behind a transfer-heavy ATS site may die on cycle (DoD and event count) while float voltage still looks normal. Log both clocks in CMMS:
- Calendar age since install / last full refresh
- Significant discharge events per year (and approximate DoD)
If the site exceeds roughly 10–20 significant discharges per year, treat cycle wear as co-equal with calendar age when setting EOL margin. Chemistry trade-offs: lead acid vs lithium UPS battery.
Planning formula: EOL Ah margin #
Planning Ah ≈ Day-one Ah ÷ EOL_fraction
Or equivalently: size day-one Ah for the energy need, then require aging margin so degraded strings still meet contract minutes at EOL.
| Policy | EOL_fraction | Meaning | Typical use |
|---|---|---|---|
| Aggressive | 0.90 | Thin margin | Rare for critical IT |
| Standard | 0.80 | Common industrial planning | Most plants |
| Conservative | 0.70 | Hot rooms or long horizons | Electrical rooms >30°C |
How to use the fraction in tools:
- Compute day-one Ah in UPS Battery Calculator for design kW and minutes.
- Divide by EOL_fraction to get the catalog floor you order.
- For year-N risk and acceptance, enter effective Ah (catalog × EOL_fraction, or measured remaining capacity) in UPS Runtime Calculator—not the nameplate alone.
Worked example A — IT rack: 2 kW · 30 min · 48 V #
Assumptions: η ≈ 0.85, DC bus 48 V, contract 30 minutes at 2 kW critical load.
- Day-one Ah ≈ 24.5 Ah (energy math from the battery calculator)
- At 80% EOL: planning Ah ≈ 24.5 / 0.80 ≈ 30.6 Ah catalog floor
- Confirm ≥30 min at design kW in UPS Runtime Calculator using installed Ah after aging, not catalog day-one
Year-4 screening: If usable capacity is ~80% and impedance has risen, minutes often sit in a ~24–28 min band before hard cutoff—enough to fail a 30 min witnessed test even though LEDs still read float OK.
Worked example B — hot room + same load #
Keep 80% EOL and shorten calendar refresh (e.g. 3–4 years instead of 5–7) when the battery cabinet averages ~10°C above the 20–25°C reference. Arrhenius-style life halving is a rule of thumb—confirm on OEM charts and pair with temperature derating.
Re-run runtime at year-3 impedance assumptions: minutes often collapse before float alarms. Do not “buy more Ah forever” without cooling—heat multiplies both calendar fade and impedance rise.
Worked example C — network closet / CCTV bridge (short minutes) #
Target: 0.4 kW · 15 min bridge for a PoE stack and NVR (see scenario sizing on UPS for CCTV and UPS for network closet).
- Day-one Ah ≈ 2.0–2.5 Ah class at 48 V (order of magnitude—run the calculator for exact η and voltage)
- With EOL 0.80: order ~25% more Ah than day-one
- Aging risk here is false confidence: short bridges look fine on self-test until impedance rise cuts the last 3–5 minutes you need for graceful shutdown
After any string refresh, re-verify minutes at measured closet load in the runtime calculator.
Impedance trends vs self-tests #
Built-in UPS self-tests are short. They rarely reproduce full critical load or the voltage sag of a high-impedance aged string. Trend internal ohmic / impedance:
- Baseline after burn-in (new string)
- Semi-annual points in CMMS (same meter, same temperature window if possible)
- Replace near 130–150% of baseline impedance or on failed load-test minutes—whichever comes first
Photograph terminal conditions and log ambient temperature with each reading—cold and hot cabinets skew ohmic numbers. Full procedures: UPS Battery Maintenance.
Mixing new jars into old strings #
Replacing only the “worst” blocks while leaving aged companions in series creates current hogging and uneven DoD. Runtime can fall non-linearly—worse than the average Ah of the mixed string suggests. Prefer full string refresh unless OEM guidance and impedance maps explicitly allow selective replacement.
Worked comparison: day-one vs year-4 minutes #
| Case | Ah used in model | Load | Approx. screening minutes |
|---|---|---|---|
| Commissioning | 50 Ah @ 48 V | 2 kW | ~40+ min (before derates) |
| Year 4 (80% usable) | 40 Ah effective | 2 kW | ~20% less energy → ~32 min class |
| Year 4 + high impedance | Earlier DC cutoff | 2 kW | Can undershoot contract 30 min |
Exact minutes need OEM discharge curves and the runtime calculator—this table is directional for stakeholder conversations, not a stamped acceptance certificate.
Who owns the aging budget? #
| Role | Decision | Artifact |
|---|---|---|
| Engineering | EOL_fraction, contract minutes | Battery standard + calculator screenshots |
| Operations | Impedance / load-test cadence | CMMS schedule |
| Finance | Refresh year vs CapEx | EOL Ah vs day-one delta |
| Facilities | Ambient control | Temp logs → derating + refresh |
Aging is not only a battery SKU problem—it is a runtime SLA problem. Document the SLA as minutes at X kW at EOL, then reverse into Ah and refresh cadence.
Common mistakes #
- Sizing to day-one Ah with no EOL margin.
- Trusting green float LEDs while impedance has already climbed.
- Skipping load tests after string refresh or major load adds.
- Mixing new jars into old strings—resistance imbalance kills runtime non-linearly.
- Ignoring temperature history—hot rooms need both EOL margin and earlier refresh (see temperature derating).
- Using self-test green as proof of contract minutes.
Next steps #
- Write EOL fraction (e.g. 0.80) and contract minutes @ kW into your battery standard.
- Size Ah in UPS Battery Calculator with aging margin; verify minutes in UPS Runtime Calculator at effective Ah.
- Schedule impedance trending and load tests per UPS Battery Maintenance.
- If cabinets run hot, apply temperature derating before you only add Ah.
- Return to the UPS calculator hub for the load → capacity → runtime → battery workflow.
How much runtime does UPS battery aging typically remove?
Often 15–30% of minutes by mid-life under heat and float aging—site-specific. Plan with EOL Ah (e.g. 80% of day-one) and confirm with a loaded runtime test, not self-test alone.
Should I enter aged Ah or day-one Ah in the runtime calculator?
For acceptance and year-N risk, enter the effective Ah you expect at EOL (or raise the safety factor). For ordering, size catalog Ah above that floor.
Does lithium age the same way as VRLA?
No. Lithium uses BMS SoH and cycle counts; VRLA leans on impedance and calendar life. Chemistry choice: lead acid vs lithium UPS battery.
When should we replace strings for aging alone?
When impedance or load-test minutes cross policy or calendar age hits OEM max—even if float looks normal. Do not wait for the first failed outage.
How is this different from UPS battery maintenance?
This page explains why minutes shrink and how to set EOL margin. Maintenance covers what to measure and when (impedance, load tests, CMMS).
Can I replace only the weak blocks in a UPS string?
Usually avoid selective jar swaps unless OEM guidance and impedance maps allow it. Mixed new/old blocks create imbalance and non-linear runtime loss—prefer full string refresh.