Introduction #

When this guide fits: You operate or maintain VRLA (or similar) strings behind UPS and need a replacement and test discipline tied to runtime risk—including when to buy new batteries.

When it is not suitable: You are designing a new DC protection and grounding scheme for a substation—follow utility standards and PE-directed models instead of this UPS-focused overview.

The UPS inverter is only as good as the DC bus behind it. Valve-regulated lead-acid (VRLA) strings remain the default in many plants because they are predictable and well understood—but they age silently. Capacity fades from grid overcharging, stratification, elevated temperature, and shallow cycling. This guide summarizes how maintenance teams test batteries, when to replace them, and how to stay safe while working on live DC systems.

Think of batteries as wear items with hidden state: the charger can look “green” while energy available for your stated minutes is already gone. The sections below turn that hidden state into trendlines, tests, and purchase triggers your CMMS can enforce.

Pair this guide with UPS Battery Calculator and UPS Runtime Calculator so replacement strings still meet the original runtime acceptance test.

Operators watch output alarms; batteries fail silently until internal resistance rises enough that terminal voltage collapses under load. Self-tests often use partial discharge and may not reproduce data-center shutdown stacks. Treat minutes as a distribution: new string, mid-life, and EOL scenarios belong in the same CMMS record so finance sees refresh coming, not only engineering.

What to measure (and how often) #

Test Typical frequency What it tells you
Visual inspection (leaks, swelling, corrosion) Monthly Mechanical failure precursors
Float voltage per string/block Quarterly Charger alignment, weak cells
Inter-cell/terminal torque check Annual Resistance heating
Internal ohmic / impedance test Semi-annual to annual Trending toward end-of-life
Load test (constant current or AC load bank on UPS) Per OEM or after events Remaining actual minutes

Trending beats spot checks. A cell that jumps 20–30% in impedance versus its baseline is often near failure even if float voltage looks fine. Plot monthly points; annual dots hide acceleration in the last quarter before failure when teams skip winter windows during busy shutdown seasons every fiscal year.

Sample UPS maintenance schedule #

Use this PM calendar as a starting template—align intervals to your OEM manual, insurance rider, and criticality tier. The goal is one CMMS record per UPS asset with next due dates, not ad-hoc heroics before holidays.

Interval UPS / charger Batteries Documentation
Monthly Review alarms, fans, bypass indicator Visual inspection (leaks, swelling, corrosion) CMMS ticket + photo if anomaly
Quarterly Log load vs nameplate rating Float voltage per string/block Trend sheet attached to asset
Semi-annual Firmware audit if site policy requires Internal ohmic / impedance test CSV export to string ID
Annual Torque / IR scan on terminations Load test per OEM or contract Signed report + runtime re-check
Event-driven After outage, transfer, or thermal alarm Impedance + visual before return to service Incident work order linked to asset

After any string replacement or load change, re-run UPS Battery Calculator and UPS Runtime Calculator so the schedule targets the current protected kW and Ah basis. For multi-string layouts, use the battery strings and banks section when block count changes.

More workflow context: UPS calculator hub (load → capacity → runtime → battery → bank).

Temperature discipline #

For VRLA, Arrhenius-style rules of thumb: every 10°C above 20°C can roughly halve calendar life. If the battery cabinet shares a hot motor room without dedicated cooling, derate the design replacement interval and re-run UPS Runtime Calculator at elevated resistance assumptions.

Battery life calendar and cycle limits #

Calendar life (years on float) and cycle life (event count) are different budgets—both can end a string before impedance alarms trip.

Chemistry Typical calendar planning Cycle note
VRLA (AGM/gel) 3–5 years industrial; 5–7 years controlled IT rooms Frequent shallow outages add cycle wear
High-rate VRLA Often shorter calendar at same temp Good for short bridges, watch replacement interval
Lithium (site-approved) OEM-dependent; often longer calendar BMS tracks cycles; different disposal path

Rule of thumb: If the site exceeds 10–20 significant discharge events per year, treat cycle wear as the driver—not only calendar age. After any string replacement, update baseline impedance and re-run UPS Runtime Calculator at design kW.

For replacement procurement, size Ah in UPS Battery Calculator and lay out blocks in the strings and banks section when block count changes.

DoD and end-of-life margin #

DoD is the fraction of nameplate energy removed during an event:

DoD ≈ usable Wh removed ÷ nameplate Wh × 100%

Many teams embed DoD indirectly via aging margin, efficiency η < 1.0, and OEM end-of-discharge voltage—not 100% of catalog Ah.

Planning stance Typical usable DoD (VRLA UPS) Effect on life
Conservative IT 50–60% Longer calendar life
Standard bridge 60–80% Common for 10–15 min designs
Aggressive (discouraged) >80% Shortens life; needs OEM sign-off

Example: 100 Ah string, plan 60% DoD → budget 60 Ah usable per event before recharge—not 100 Ah.

Worked EOL margin (2 kW · 30 min · 48 V · η 0.85): day-one Ah ≈ 24.5 Ah; add 20% aging → 29.4 Ah planning floor; at 60% DoD catalog ≈ 49 Ah per string (OEM tables override). Confirm ≥30 min in UPS Runtime Calculator at design kW and log the result in CMMS.

When to replace #

Replace a string proactively when:

  • Impedance trends cross the vendor threshold or internal policy (many sites use 130–150% of baseline).
  • A load test cannot meet the required runtime at end-of-life voltage—even if the UPS still “passes” a brief self-test.
  • Physical bulging, terminal corrosion, or thermal hotspots appear on IR inspection.

After any nearby fire, flood, or seismic event, perform impedance and load tests before returning the bus to service.

Root causes teams mislabel as “charger bugs” #

Loose intercell links mimic rectifier ripple. Corroded shunts change sense wiring and drive wrong float. Before swapping expensive charger boards, torque and clean first, then re-measure ripple at known load steps. Document before/after scope captures for the vendor case file.

Safe work practices #

  • Use insulated tools, arc-rated PPE per site policy, and LOTO on all AC feeds to the UPS and maintenance bypass.
  • Short-circuit current on large parallel strings can be extremely high; use rated disconnects and training.
  • Label polarity and torque values at installation so future crews cannot mirror-wire a replacement tray.

Receipt inspection and baseline (day zero) #

Before burn-in, photograph serial numbers, terminal orientation, and inter-tier spacing. Record open-circuit voltage per block after stabilization and compare to OEM ranges. Baseline impedance after temperature equalization gives the trend anchor—do not skip because “the UPS passed self-test.” File the baseline under the string ID, not only the UPS asset ID, because strings rotate independently on some modular systems.

Float, equalize, and charger discipline #

Float voltage that wanders high cooks jars; low float invites sulfation in some designs. Follow the OEM chart for temperature-compensated float where available. Equalize cycles (where permitted) are not “free maintenance”—they consume water in flooded cells and stress VRLA if misapplied. Log equalize events beside room temperature so you can correlate capacity dips later.

Symptom Charger-side check Battery-side check
High float current Setpoints, calibration Internal short risk
Ripple on DC bus Rectifier health Loose intercell links
Uneven block temps Sharing harness Weak cell

Load testing without drama #

Use a dedicated load bank procedure when possible rather than “inventing” a test by opening random plant loads. Define stop conditions: minimum DC voltage, maximum cell temperature, and maximum test duration. Assign one radio net leader so E-stop is unambiguous. After the test, recharge to full before returning to service—returning with a depleted string is how false alarms multiply.

Worked example — impedance trend (illustrative) #

Baseline: 2.10 mΩ average on a 120-block string. Year 3: 2.45 mΩ (+17%). Year 4: 2.90 mΩ (+38% vs baseline). Policy says replace at +35% or failed load test—schedule replacement now rather than waiting for a holiday outage surprise.

Try our UPS Runtime Calculator with elevated internal resistance assumptions to see how minutes collapse at year-end.

Spill containment, seismic, and housekeeping #

VRLA jars can crack from forklift bumps or seismic motion. Spill kits belong inside the room, not in a distant warehouse. Cable dress should not lift terminals—strain shows up first as heat on IR scans. Keep debris out of air intakes on cabinet fans; hot chargers age adjacent jars unevenly.

Battery maintenance loop: inspect, trend, load testInspect→Trend Ω→Load testDocument every step in CMMS with photos.

Disposal, hazmat, and vendor returns #

Used batteries are regulated in many jurisdictions. Keep manifests aligned to serial lists. If a vendor offers credit for cores, verify who owns freight and insurance during return. Never stack unknown chemistries on the same skid.

UPS decommissioning and end-of-life planning #

Decommissioning checklist:

  1. LOTO all AC feeds and maintenance bypass paths; verify zero energy on DC bus with approved meters.
  2. Export SNMP/Modbus logs and alarm history for asset records.
  3. Drain or stabilize batteries per OEM; use licensed hazmat haulers for VRLA/lithium.
  4. Remove network cards from VLAN ACLs; revoke SNMP credentials.
  5. Update owner single-line and CMMS to retired status.

Warranty and lifecycle: Track inverter warranty, battery pro-rata, and extended service contracts in the same CMMS record as impedance CSVs. Plan refresh at 70–80% of design calendar life in hot rooms—not only when alarms trip. Finance should see battery Capex as a recurring line item, not a surprise capital project.

For maintenance bypass procedures during removal, follow the site MOP in Complete UPS Sizing Guide (Step 5 and bypass deep dive)—never hot-work a parallel bus without vendor sign-off.

Insurance and incident documentation #

After thermal events, preserve BMS logs for lithium systems and charger logs for VRLA. Insurers often ask for proof of scheduled tests—CMMS exports beat memory.

Training: what good looks like #

Technicians should demonstrate LOTO, insulated tool selection, and first-responder steps for DC arc risk. Pair senior mentors with contractors on the first string swap. Quiz on polarity mistakes—those cause catastrophic weekends.

Try our UPS Battery Calculator when replacement strings change Ah or count so runtime acceptance matches the original MOP.

Parallel strings, current sharing, and “one weak jar” #

When multiple strings feed a common DC bus, sharing harnesses and fuses must be symmetric. A weak jar raises float current on its neighbors, accelerating imbalance. If you see one string warmer than its twin, investigate connector torque and inter-string voltage offsets before blaming ambient alone.

Interpreting infrared without wishful thinking #

IR photos are useful, but emissivity on plastic cases can lie. Compare delta-T jar-to-jar on the same rack face under similar charge states. Hot terminals often mean resistance; hot case mid-body may mean internal degradation—flag both, but do not conflate root causes.

Alarms you should not silence permanently #

High float current, charger faults, and temperature rate-of-change alarms deserve tickets with owners. If operations masks alarms during busy season, you will discover end-of-life during the first real outage. Write a policy: which alarms may be snoozed, for how long, and who approves.

Spare parts and shelf life #

Spare jars age on the shelf too—track date codes and rotation into service. Torque hardware kits should live with the UPS asset. Fuse types should be non-substitutable without engineering review because AIC and speed matter on DC branches.

Second worked example — minutes collapse (illustrative) #

Design: 10 minutes at 200 kW on a new string. Year 5: internal resistance model suggests 18% longer voltage sag under same kW. Effective minutes drop near 8 without any SNMP warning if thresholds are naive. Update MOP minimum minutes or refresh early.

Vendor PM contracts: read the fine print #

Some contracts count a visual walk as a “test.” Define deliverables: impedance CSV, photos, torque sheet, and load test curve when required. Tie payments to evidence, not checkboxes.

Winter versus summer test windows #

Cold rooms slow chemical kinetics; hot rooms accelerate aging but can mask low float symptoms if chargers temperature-compensate aggressively. When comparing year-to-year tests, annotate ambient and charger mode so you do not misread a seasonal shift as failure.

Handover package checklist #

Artifact Owner Frequency
Impedance CSV Electrical Semi-annual
Load test report Contractor Per contract
Torque log Mechanical Annual
Firmware rev IT/OT Each change

When to escalate to the OEM early #

Escalate when parallel strings show divergent impedance slopes, when BMS logs show cell imbalance acceleration on lithium, or when charger ripple doubles versus commissioning. Waiting for smoke costs weeks of lead time.

Closing discipline #

Good battery programs are boring: predictable dates, trending graphs, and few surprises during outages. If your program feels exciting, you are probably funding heroics instead of maintenance—and heroics rarely scale across ten sites.

Try our UPS Load Calculator when protected kW changes so runtime tests target the right load step.

Next steps you should take #

  1. Schedule impedance trending on the next maintenance window; baseline any new strings after burn-in.
  2. Re-run UPS Runtime Calculator after any string replacement to confirm minutes at design load.
  3. File reports under the UPS asset ID in your CMMS with next review date.
  4. Link photos and CSV exports to the same work order so auditors see one chain of evidence.

More UPS tools: UPS calculator hub.

What DoD should I use for a 15-minute UPS bridge?

Many VRLA IT designs plan 60–80% usable energy for a 15-minute event—then add aging margin so EOL strings still meet 15 minutes. Confirm with OEM discharge curves, not rules of thumb alone.

How do cycles affect replacement if the battery rarely discharges?

Calendar life still applies—float and temperature age VRLA even without outages. Frequent self-tests add cycle wear; log them in CMMS.

What should a UPS maintenance schedule include?

At minimum: monthly visual checks, quarterly float voltage logging, semi-annual impedance trending, and annual torque/IR and load tests where policy requires. Tie each task to a CMMS due date, attach CSV/photos, and re-run runtime acceptance after string replacement or load changes.

Are UPS “self-tests” enough?

Built-in self-tests are short and may not reveal capacity loss. Schedule periodic impedance and occasional load tests for critical buses.

Can I replace only one jar in a string?

Usually no for VRLA strings: a new cell in an old string becomes the electrical weak point and mismatches charging. Follow OEM guidance; most sites replace entire strings.

What about lithium-ion UPS batteries?

Different BMS alarms, thermal runaway mitigation, and shipping rules. Treat as a separate program—do not assume VRLA checklists transfer.

How often should we load-test a critical string?

Follow OEM and insurance requirements; many critical sites use annual or biennial tests with semi-annual impedance trending between them.

What torque mistakes cause fires?

Under-torque creates resistance heat; over-torque cracks posts. Use the OEM chart and a calibrated wrench.

How do I perform a UPS battery discharge or load test?

Use a defined load bank or approved UPS load test: set stop conditions (minimum DC voltage, max duration, max cell temperature), assign one radio net leader, document start/end voltages, and recharge to full before returning to service. Follow OEM and insurer intervals—often annual or biennial for critical strings with semi-annual impedance between tests.

What is a safe UPS firmware update policy?

Schedule firmware during planned maintenance windows with bypass/MOP approved, test on a non-production unit when possible, snapshot config, and verify SNMP/NMS traps after reboot. Never push firmware during storm season production peaks without rollback plan.

How should operators respond to common UPS alarm codes?

Treat on battery, low battery, overload, fan fault, and bypass active as actionable—log code, load %, and estimated runtime; execute load shed or generator MOP per runbook. Do not permanently silence alarms without a CMMS ticket and owner.