Datacenter
UPS battery maintenance and testing field guide for data centers
Commission, test, and maintain UPS battery strings: the capacity discharge test, impedance trending, float and temperature control, and the records that prove the runtime before the outage does.
Direct answer
A UPS battery is the stored energy that carries the critical load when the source fails, and it is the part of the UPS most likely to fail when called, quietly, between tests. Capacity discharge testing and impedance trending are what prove the runtime is real before an outage does. The manufacturer and project spec control acceptance.
Key takeaways
- Capacity discharge testing and impedance trending are the only proof UPS runtime is real; float voltage says almost nothing about capacity.
- Replace a lead-acid string below 80 percent of rated capacity (recognized end of life); 90 percent or more is healthy.
- IEEE 1188 covers VRLA, IEEE 450 covers flooded, IEEE 1106 covers nickel-cadmium; lithium follows the manufacturer and BMS procedure.
- Lead-acid service life roughly halves for every 10 degrees C sustained above the rated 25 degrees C.
- IEEE 1188 treats an ohmic deviation over 20 percent from the commissioning baseline as cause for investigation or replacement.
The battery is the weak link in the UPS
The UPS battery is the stored energy that carries the critical load through the gap when the source fails. It is also the part of the whole power chain most likely to let you down when it is finally called, and it fails in a way nothing else in the room does: quietly. A capacitor bulges. A fan rattles. A breaker trips. A battery just gets weaker, cell by cell, with no symptom on the front panel until the day the utility drops and the string cannot make its minutes.
That is why this guide is about testing and trending, not about hoping. The UPS itself is covered in the commissioning hold-points guide, the inverter, the static bypass, the transfers. The battery is its own discipline. The rectifier can be perfect and the inverter flawless, and if the string behind them has lost a third of its capacity to heat and age, the autonomy the owner paid for is a number on a data sheet and nothing more.
The hard part is that a battery looks fine right up until it is not. A string sitting on float reads its full voltage, draws a trickle of current, and gives every appearance of health. Voltage on float tells you almost nothing about capacity. The only things that tell you whether the runtime is real are the discharge test that measures it directly and the ohmic trend that warns you a jar is going before the discharge proves it. Skip both and the outage becomes your capacity test.
VRLA, flooded VLA, and lithium-ion: the three batteries you commission
Three chemistries cover almost all UPS battery work, and they are not interchangeable in how you commission or live with them. VRLA, valve-regulated lead-acid, is the data center default: sealed, no watering, lower first cost, and it drops into a cabinet next to the UPS. The trade-off is life and temperament. VRLA is commonly rated around 3 to 5 years of real service life in UPS duty, sometimes up to 10 on premium pure-lead designs, it is sensitive to heat, and it can go into thermal runaway if it is overcharged or cooked.
Flooded lead-acid, the vented or VLA cell, is the long-life option that buys its life with maintenance. A flooded string can run 15 to 20 years, you can see the plates, and a single failed cell is replaceable, but the cells gas on charge, lose water, and need periodic watering, equalize charges, and a vented room with the life-safety stack that goes with it. Its footprint and upkeep are why it has faded from new builds even though it outlives everything else.
Lithium-ion, usually lithium iron phosphate in UPS service for its thermal stability, is the shift in the market. It lasts longer, often 10 to 15 years, holds roughly three to five times the energy of VRLA in the same space and weight, recharges faster, and tolerates a warmer room. It costs more up front and it arrives with a battery management system that is part of the protection, not an accessory. Pick the chemistry against the life, the floor space, the room you have, and the budget, and commission each one to its own rules.
| Attribute | VRLA (sealed) | Flooded VLA | Lithium-ion (LFP) |
|---|---|---|---|
| Typical service life | About 3 to 5 years, up to ~10 premium | Often 15 to 20 years | Often 10 to 15 years |
| Maintenance | Sealed, no watering, ohmic and capacity tests | Watering, equalize, specific gravity, vented room | BMS-managed, mostly visual and data review |
| Footprint and weight | Baseline for the room | Largest for the energy | About 1/3 to 1/5 of VRLA |
| First cost | Lowest | High | Highest |
| Key risk to manage | Heat and thermal runaway | Hydrogen, electrolyte, watering | Cell-level fault, fire listing |
How do you test a UPS battery's capacity?
You test capacity with a discharge, also called a load or capacity test, and there is no substitute for it. Charge the string fully, let it stabilize on float, then discharge it at a defined constant current or constant power to a specified end-of-discharge voltage, and time it. The delivered capacity is what you measure against the rating. A test run at the published rate that reaches the end voltage in the rated time is a string at 100 percent. One that hits the end voltage early has lost capacity in proportion to how early it got there.
The defined rate matters as much as the result. A battery rated for a certain runtime at a given discharge rate will deliver a different number at a different rate, because lead-acid capacity falls off as you pull harder, so you discharge at the rate the test calls for and correct the result to the reference temperature. The IEEE recommended practices frame the work by chemistry. IEEE 1188 covers maintenance, testing, and replacement of VRLA. IEEE 450 covers the same for vented lead-acid (flooded) cells. IEEE 1106 covers vented nickel-cadmium. Lithium follows the manufacturer's and BMS maker's procedure. Cite the one that matches the chemistry on the floor.
An acceptance capacity test belongs at commissioning, before the string ever carries a real load, because it is the baseline every later test is judged against and the only proof the autonomy you bought is real. Run it with a calibrated load bank, record the full discharge curve and not just the endpoint, and log the battery and room temperature it was taken under. A discharge done between roughly 65 and 90 degrees F, corrected to the reference, keeps the result comparable to the next one. The autonomy verification covered later in this guide rides on this same test, run at the design load.
What capacity does a UPS battery have to pass?
A healthy stationary battery delivers 90 percent or more of its rated capacity, and the recognized end of life is 80 percent. Below 80 percent of rated capacity, the cell is considered worn out and due for replacement, because lead-acid capacity does not coast down gently from there. Once a string drops through 80 percent it tends to fall off quickly, so the threshold is set to catch the battery while it can still be replaced on a schedule instead of after it fails a load.
The intermediate numbers drive the testing interval, not just the verdict. Common practice from IEEE 1188 and 450 is to capacity test when a string has reached about 85 percent of its expected service life, or when capacity has dropped more than 10 percent from the previous test, or has fallen below about 90 percent of rating, and then to test annually from there. A result in the 80s is not a pass to celebrate. It is a string on the clock that needs a date set for replacement.
These thresholds are recognized practice, not a one-size number. The exact acceptance percentage, the discharge rate, and the end-of-discharge voltage come from the manufacturer's data and the project specification, and the warranty often ties to them. Hold to the contract figures first, fall back to the IEEE practice and the manufacturer's published values where the spec is silent, and treat 80 percent as the line you replace at, not the line you run to.
| Capacity vs rating | What it means | Action |
|---|---|---|
| 90 percent or more | Healthy string | Continue normal cadence |
| 80 to 90 percent | Aging, accelerated trend likely | Test annually, plan replacement |
| Below 80 percent | Recognized end of life | Replace the string |
| Drop over 10 percent since last test | Falling faster than age alone | Investigate, shorten interval |
What is a battery impedance test?
A battery impedance test measures the internal ohmic value of each cell or jar, and it is the predictive method that warns you of a weak cell before a discharge test would. The instrument reads impedance, conductance, or resistance, all forms of the same ohmic measurement, on every cell in the string while it sits on float. The point is not the single reading. It is the trend of that reading against a baseline taken when the battery was new and healthy.
Internal resistance rises as a lead-acid cell ages and dries, so a jar whose ohmic value has climbed well above its baseline and above its neighbors is the jar that will drag the string down on the next discharge. IEEE 1188 frames this directly: a recent edition treats an ohmic deviation greater than 20 percent from the cell's commissioning baseline as cause for replacement or immediate investigation, and a rise on the order of 30 to 50 percent in an individual cell is a strong flag that its capacity has fallen below 80 percent. The numbers vary by instrument and cell type, so trend against your own baseline, not someone else's table.
Take the baseline right. Establish it on a settled battery, commonly a few weeks after the last discharge and after the string has been on float long enough to stabilize, and take it with the same instrument and the same lead placement you will use forever after, because ohmic readings are method-sensitive. The value of the impedance test is that it is fast, non-intrusive, and can run on a live string without discharging it, so it bridges the gap between the annual capacity tests and catches the cell that fails in between. It does not replace the capacity test. It tells you when to worry before you run one.
Float, boost, and the charger that has to be temperature-compensated
Float is the steady voltage the charger holds on the battery to keep it fully charged and offset self-discharge without overcharging it. For VRLA, float commonly lands around 2.25 to 2.30 volts per cell at 25 degrees C, set to the manufacturer's curve; flooded cells float a touch lower and get a periodic equalize or boost charge, a deliberately higher voltage that brings up lagging cells and stirs the electrolyte. Lithium does not float in the lead-acid sense. The charger runs a constant-current, constant-voltage profile to a target and then backs off, and the BMS balances the cells, so there is no equalize charge and no float voltage to chase.
The float setpoint is not a fixed number across temperature, and this is where strings get cooked. As a lead-acid battery warms, the same float voltage drives more current into it, so the charger has to be temperature-compensated: the float is lowered as battery temperature rises and raised as it falls, on the order of a few millivolts per cell per degree C, around 3 mV per cell per degree C as a common figure, against the manufacturer's spec. The sensor has to read the battery, not the room, and it has to be on the battery, working, and trusted by the charger.
Get this wrong and the failure is in the title. A hot string on a fixed, uncompensated float overcharges, heats further, and walks itself toward thermal runaway. A float set too low chronically undercharges and sulfates the plates, quietly stealing capacity you will not see until the discharge test. Confirm the float and equalize values against the manufacturer's curve at the measured battery temperature, and confirm the temperature compensation is active and reading the right sensor. That single check prevents both ends of the failure.
What causes VRLA thermal runaway?
Thermal runaway is a self-feeding loop: a cell on charge gets hot, the heat lowers its internal resistance and raises the current it draws at the same voltage, the higher current makes it hotter, and around it goes until the cell is destroyed and venting. It is worst in VRLA because the cells are sealed and packed tight, with little airflow between jars to carry heat away, so a hot cell stays hot and pulls its neighbors with it. The charging current can climb dramatically for a small rise in per-cell voltage, which is why an uncompensated float on a warm string is the classic trigger.
The mechanism is overcharge plus heat with nowhere to go. A fixed float voltage held while the room or the battery runs hot, a charger with no working temperature compensation, a cell that has lost the ability to recombine its gas, or a room that does not move air, any of these tips the balance. The sealed VRLA design recombines oxygen internally under normal charging, but in runaway it gasses faster than it can recombine and vents hydrogen through its one-way relief valve, and a charged string can keep venting until something cuts the charge voltage or the system is shut down.
Ventilation and float control are the two defenses, and they are not optional. Keep the room temperature down and moving air so heat leaves the cells, keep the charger temperature-compensated so the float drops as the battery warms, and watch per-string current and temperature so a string that starts pulling more current than its siblings raises an alarm before it runs away. A battery monitoring system that trends cell temperature and string current is the early warning here. The lithium contrast is in the BESS guide, where the BMS is the layer that interrupts a cell-level fault before it propagates.
How does temperature affect battery life?
Heat is the single biggest thief of battery life, and the rule of thumb is brutal: for lead-acid, service life roughly halves for every 10 degrees C (about 15 to 18 degrees F) of sustained operating temperature above the rated 25 degrees C (77 degrees F). The chemistry follows the Arrhenius relationship, where reaction rates climb exponentially with temperature, so a string rated for 5 years at 77 degrees F can be down to about half that at 95 degrees F if it lives there. The number is a guide, not a contract, and it tends to run optimistic for the hard, short, high-current duty a UPS battery actually sees.
That makes battery room temperature a design and a maintenance item, not a comfort setting. The room sized and cooled to hold the battery near 25 degrees C is the room whose batteries make their rated life. The room that drifts warm because the cooling is undersized, a vent is blocked, or the batteries sit too close to the UPS heat is the room that replaces strings years early and never knows why. Lithium tolerates a warmer ambient than lead-acid, but it is not immune, and its own life and warranty track temperature too.
The field lesson is to treat the thermometer as a battery instrument. Trend room and battery temperature, find and fix the hot spots, and remember that a string that barely makes its capacity at a cool acceptance test has no margin once the room warms and the cells age. Build the headroom into the cooling, because you cannot get the life back after you have baked it out.
Intercell connections, torque, and the infrared scan
Every joint between cells, between tiers, and at the string terminals is a place the current has to cross, and a bad one is a failure that shows up at the worst moment. Intercell connection resistance gets measured against the manufacturer's limit with a micro-ohmmeter and the connections torqued to the manufacturer's value in the specified sequence, because a high-resistance link runs hot under discharge current, drops voltage exactly where you need it, and can be the reason a healthy string still underperforms on a load test. This is the check crews rush, and it is one of the most common findings on a string that fails for no obvious reason.
Torque alone is not proof, because a connection can read the right torque and still be high resistance from corrosion or oxidation between the surfaces. Measure the resistance, compare it cell to cell, and apply the manufacturer's anti-oxidation compound where they call for it. Put a witness mark on each joint after torquing so a later check can tell you if one moved.
Infrared closes the loop, and it has to be done under load. A loose or corroded joint only shows its heat when current flows through it, so scan the intercell links and terminals with the string discharging or carrying real current, and any joint reading hot against its neighbors gets cleaned, re-torqued, and re-scanned. A thermal scan on a floating string at trickle current proves very little. The same infrared discipline applies across the gear: scan connections energized and loaded, not cold, because the heat is the evidence and the heat needs current to exist.
Hydrogen, ventilation, and the battery room
Lead-acid batteries off-gas hydrogen on charge, and hydrogen becomes flammable in air at around 4 percent by volume, so the battery room's ventilation is a life-safety system, not a comfort one. Flooded cells gas the most and continuously, which is the main reason they live in a dedicated, vented room. VRLA gasses far less in normal recombinant operation, but it vents hydrogen during overcharge and thermal runaway, so it is not exempt from the ventilation question. The ventilation has to move enough air to keep hydrogen well below the lower explosive limit under the worst-case charging the room will see.
The IEEE 1635 and ASHRAE Guideline 21 practice covers ventilation and thermal management for stationary battery installations across chemistries, and the fire code, the NEC storage-battery provisions in Article 480, and the authority having jurisdiction set what is enforced. Confirm the ventilation actually runs and moves air, not just that a fan exists, and confirm any hydrogen detection and its interlock to the ventilation operates on a real signal.
Flooded rooms carry more of the life-safety stack because the electrolyte is liquid sulfuric acid. Confirm the eyewash and emergency shower are present, plumbed, and within reach, the spill containment and acid neutralization kit match the electrolyte volume, and the personal protective equipment for handling electrolyte is on hand and used. The DC hazard rides over all of it. A charged string holds full voltage at its terminals with every breaker open and can deliver a fault current that vaporizes a dropped wrench, so treat the terminals as live, use insulated tools, and follow the arc-flash and shock study the same way the UPS commissioning guide does.
Lithium-ion: the BMS, state of health, and the fire question
Lithium-ion changes the maintenance model because the battery polices itself. Every lithium UPS battery comes with a battery management system that monitors voltage, temperature, and current at the cell or module level, balances the cells, and trips the battery off line in milliseconds when a cell leaves its safe band. There is no equalize charge and no per-cell float to set by hand, because the BMS owns the cell management. Commissioning a lithium plant is largely commissioning that BMS: confirming addressing and communication, confirming the telemetry is sane, and proving that a forced limit actually opens the battery breaker.
State of health is the lithium version of the capacity number. The BMS reports state of charge, how full the battery is now, and state of health, its capacity now against when it was new, and much of lithium maintenance is reviewing that data and a yearly visual of the racks rather than running a full discharge. Trust the BMS, but verify it against the UPS and any monitoring at commissioning, because the UPS charger and the BMS both think they own the battery and their limits have to agree. A capacity check still belongs in the program where the spec or warranty calls for one.
The fire question is what makes lithium its own discipline. A failing lithium cell can vent flammable gas and go into thermal runaway that propagates cell to cell, which is why lithium energy storage is generally listed to UL 9540 as a system and evaluated by the UL 9540A fire-propagation test method, and why NFPA 855 governs the installation, spacing, detection, and ventilation where it applies, commonly above a 20 kWh threshold. The grid-scale and room-scale version of all of this, including UL 9540A and NFPA 855, is worked in detail in the BESS commissioning guide. For a UPS lithium cabinet, confirm the listing, the fire detection and protection the design calls for, and that the installation matches what the 9540A data and the listing assumed.
How do you verify UPS battery runtime?
Runtime, the autonomy, is the number the owner actually bought, and it only counts at the design load. Verify it by discharging the string at the real design load the room will see, timing it from full charge to the end-of-discharge voltage, and comparing the measured minutes to the design autonomy. Runtime scales nonlinearly with load: a string that holds 15 minutes at half load holds far less than half that at full load, so a runtime proven at a convenient light load is not a runtime, it is a number that will not be there when the load is full.
In a data center the design autonomy is usually short on purpose, often a few minutes up to around 15, sized to bridge the gap until the generators start and accept the block rather than to run the building. That makes the margin thin by design, so a discharge that just barely makes the number on a new battery is a finding, not a pass, because every battery ages from there. The UPS-side framing of this, the transfer to battery and the generator bridge, lives in the commissioning hold-points guide; here the job is proving the string behind it delivers.
Decide between a partial and a full discharge with eyes open. A full capacity discharge to the end voltage is the definitive test but it leaves the string temporarily depleted and needs a recharge window before it can protect the load again, so it is scheduled with the UPS on bypass or with redundant capacity available. A partial or modified discharge is less disruptive and useful for trending but does not fully prove capacity. Record the whole discharge curve either way, because its shape reveals a weak cell dragging the string down early, and it is the baseline the next test trends against.
The maintenance cadence the owner inherits
When the project hands the room over, it hands over a maintenance schedule, and the recognized practice from IEEE 1188 for VRLA and IEEE 450 for flooded sets the rhythm. Monthly, the light touch: read and record string float voltage, charger output voltage and current, ambient and battery temperature, and a visual for leaks, swelling, corrosion, and cracked cases. Quarterly, add the ohmic measurement on every cell with the impedance instrument, the per-cell and string float voltages, and a check of the intercell connections. Annually, the deep one, including the capacity test under the conditions and intervals the criteria section laid out.
Flooded strings carry extra periodic work the sealed ones do not: electrolyte level and watering, specific gravity readings, and the equalize charges the cells need to stay balanced. Lithium inverts the schedule. The BMS does the continuous watching, so the human cadence is mostly a periodic visual of the racks and a review of the BMS data and trends, with a capacity check where the spec or warranty requires it.
The cadence is worth nothing without the records behind it. NFPA 70B frames maintenance intervals for electrical equipment generally, and the IEEE practices set the battery-specific ones, but the value is in the trend, not the single reading. A float voltage, an ohmic value, and a temperature logged every interval against the baseline is what turns a pile of numbers into the warning that a string is going. The exact intervals come from the manufacturer and the project's maintenance spec, so confirm them and hold to the stricter one.
Battery monitoring systems and continuous trending
A battery monitoring system is the per-cell or per-jar monitor that watches the string continuously and alarms on a problem instead of waiting for the next quarterly visit. It trends cell voltage, ohmic value, string and cell temperature, and string current, and the good ones do it around the clock, which is the difference between catching a failing cell weeks ahead and finding it on a discharge test, or worse, during an outage. For a critical room, continuous monitoring is the way the slow, quiet failure mode of a battery gets a voice.
The value is in the trend and the alarm, not the dashboard. A monitor that flags a jar whose impedance has climbed past its baseline, or a cell running hot against its neighbors, or a string pulling more current than it should, is flagging exactly the conditions that precede a capacity failure or a thermal runaway. That is an early warning a manual cadence cannot match, because the cell that fails the next capacity test usually shows it in the impedance trend months ahead if something is watching.
Monitoring supplements the tests, it does not retire them. A monitoring system reduces surprises and stretches the interval between intrusive tests in some programs, but the capacity discharge is still the only direct proof of runtime, and the monitor's own sensors and alarms get verified at commissioning so you are not trusting a blind gauge. Confirm the monitor reports to the building management system or DCIM, that its alarms actually annunciate, and that someone owns the response.
Replacement, matched jars, and disposal
When a string reaches end of life, the default is to replace the whole string, not to chase individual jars, and the reason is matching. Cells age together, and a fresh jar dropped into an old string is a mismatch: the new cell has lower resistance and higher capacity than its worn neighbors, so the string charges and discharges unevenly, the old cells get worked harder, and you have bought a weak point, not a fix. Spot-replacing a single failed jar is a stopgap to limp to a planned replacement, not a strategy.
If a jar is replaced individually as a stopgap, match it as closely as the manufacturer allows: same type, same rating, and ideally close in age or from the same production where possible, and watch it on the ohmic trend afterward. The cleaner answer on an aging string is to plan the full replacement around the capacity trend, order the matched string ahead of the failure, and schedule the swap with the UPS on bypass or redundant capacity carrying the load so the room is never unprotected during the change.
Disposal is regulated, and lead-acid and lithium are handled differently. Lead-acid batteries are one of the most recycled products there is, and they go back through the manufacturer or a certified recycler, not the dumpster. Lithium has its own handling, transport, and recycling rules, and a damaged or swollen cell is a hazard in transit, so it follows the specific procedure for the chemistry. Document the removal and the disposal chain of custody, because the record matters for both the environmental rules and the warranty.
What to document
The battery record is what lets someone who was not there judge whether the string is healthy and whether a result is defensible years later. It has to tie to the specific string by serial number and rating, and it has to carry the conditions, because a capacity result without the load and temperature it was taken under is a number that cannot be compared to anything.
Capture the string identity and chemistry, the per-cell and string float voltages, the ohmic baseline and each periodic reading, the capacity and runtime results with the discharge rate and temperature they were taken under, the battery and room temperature, and the intercell connection resistances with the torque and the witness marks. If a jar was replaced or a connection re-torqued, record the before and after so the next person sees what changed and that it was proven after the change.
| Field to record | Why it matters |
|---|---|
| String ID, chemistry, and rating | Ties the record to the specific battery |
| Float voltage, per cell and string | Shows the charger is holding the right charge |
| Cell ohmic baseline and trend | The reference that predicts a weak jar |
| Capacity test result, with rate and temperature | The direct proof of usable capacity |
| Runtime at design load | The autonomy that was actually bought |
| Battery and room temperature | Drives life and corrects every result |
| Intercell connection resistance and torque | The joints that fail under discharge |
Common mistakes
- Never taking an ohmic baseline at commissioning, so later readings have nothing to trend against.
- Running a fixed, uncompensated float so a warm string overcharges toward thermal runaway.
- Running the string past 80 percent capacity instead of replacing it at the end-of-life threshold.
- Letting the battery room run hot and replacing strings years early without knowing why.
- Torquing the intercell connections but never measuring resistance or scanning them under load.
- Skipping the acceptance capacity test, so the bought autonomy is never proven before turnover.
- Calling a string healthy off its float voltage, which says almost nothing about capacity.
- Running the runtime test at light load and reporting an autonomy the string cannot make at design load.
- Dropping a fresh jar into an old string and creating a mismatch instead of replacing the string.
- Trusting a battery monitoring system whose own sensors and alarms were never verified.
Field checklist
Want this checklist to run itself on every job — with photo proof and a signed record crews can hand the customer? That's FieldOS.
Standards and references
Several IEEE recommended practices frame UPS battery work by chemistry, and naming the right one for the point is what separates a credible record from a guess. IEEE 1188 covers maintenance, testing, and replacement of valve-regulated lead-acid (VRLA), with IEEE 1187 for VRLA installation. IEEE 450 covers maintenance, testing, and replacement of vented lead-acid (flooded) cells, with IEEE 484 for their installation. IEEE 1106 covers vented nickel-cadmium. For sizing, IEEE 485 covers sizing lead-acid and IEEE 1184 covers sizing batteries for UPS systems. Lithium leans on the manufacturer's procedure and the IEEE 1679 series for characterization and evaluation. Use the edition the project specifies.
The room and the installation bring their own standards. IEEE 1635 with ASHRAE Guideline 21 covers ventilation and thermal management of stationary battery installations across chemistries. The NEC, NFPA 70, covers the installation, with Article 480 specific to storage batteries, and NFPA 70E covers the arc-flash and shock work practices that apply to the DC string and the charger. NFPA 70B frames maintenance intervals for electrical equipment generally. For acceptance testing of the installed electrical equipment, ANSI/NETA ATS gives the field test and inspection requirements, and recent editions added coverage specific to UPS and battery systems.
Lithium energy storage carries the fire and life-safety standards. The system is generally listed to UL 9540, the separate UL 9540A test method evaluates thermal-runaway fire propagation, and NFPA 855 governs the installation of stationary energy storage where it applies. Those are worked in detail in the BESS commissioning guide. Above all of these sit the manufacturer's instructions and the project specification, which set the actual numbers, and the authority having jurisdiction, which has the final say on what is enforceable. When a standard and the spec disagree, the stricter controlling document wins, and confirm the section and edition against the adopted code before citing it on a submittal.
Units and terms
Battery work runs on a handful of units and terms, and reading the wrong one is how a result gets accepted that should not be. Capacity is in ampere-hours, or stated as a runtime in minutes at a given discharge rate, and it is judged as a percentage of rating. Float and per-cell voltage are in volts per cell, referenced to 25 degrees C. Ohmic value is in milliohms for impedance and resistance or in siemens for conductance, all forms of the same internal measurement, and what matters is its trend against the cell's own baseline.
Keep the chemistry names straight, because they decide which standard and which procedure apply. VRLA is sealed valve-regulated lead-acid, VLA is vented flooded lead-acid, and lithium-ion in UPS service is usually lithium iron phosphate. End of life is the 80 percent capacity threshold at which a lead-acid string is replaced. Thermal runaway is the self-feeding overcharge-and-heat loop that destroys a cell, and float is the steady charge voltage that, set wrong for the temperature, is what starts it.
- VRLA / VLA
- Valve-regulated (sealed) lead-acid and vented (flooded) lead-acid, the two lead-acid families with different maintenance
- Lithium-ion (LFP)
- Lithium iron phosphate, the lithium chemistry common in UPS service for its thermal stability, managed by a BMS
- Capacity test
- A timed discharge at a defined rate to an end voltage, measuring delivered capacity against rating
- Impedance / ohmic value
- The cell's internal resistance, conductance, or impedance, trended against a baseline to predict a weak cell
- Float voltage
- The steady charge voltage held on the battery, temperature-compensated for lead-acid
- Thermal runaway
- A self-feeding loop where a hot cell draws more current and gets hotter until it vents and fails
- End of life
- The 80 percent of rated capacity threshold at which a lead-acid string is replaced
FAQ
How do you test a UPS battery?
You run a capacity discharge test: charge the string fully, discharge it at a defined rate to the end-of-discharge voltage, and time it against the rating. Impedance testing on each cell trends the internal resistance between discharges. Float voltage alone proves nothing about capacity, so the discharge is the only direct proof of runtime.
When do you replace UPS batteries?
Replace a lead-acid UPS string when its capacity falls below 80 percent of rating, the recognized end of life, because capacity drops off quickly past that point. Capacity above 90 percent is healthy. Test more often once a string passes about 85 percent of its expected service life, and plan the replacement before it fails a load.
What is a battery impedance test?
A battery impedance test measures each cell's internal ohmic value, impedance, conductance, or resistance, while the string sits on float. Trended against a commissioning baseline, a rising value flags a weak jar before a discharge test would. IEEE 1188 treats an ohmic deviation over about 20 percent from baseline as cause for investigation or replacement.
How does temperature affect battery life?
Heat shortens lead-acid life sharply. Service life roughly halves for every 10 degrees C (about 15 to 18 degrees F) of sustained temperature above the rated 25 degrees C (77 degrees F), following the Arrhenius relationship. A string rated for 5 years at 77 degrees F can be down to half that in a warm room.
What causes a UPS battery to go into thermal runaway?
Thermal runaway is a self-feeding loop where a hot cell draws more charging current, gets hotter, and runs away until it vents and fails. The usual triggers are overcharge from an uncompensated float held on a warm string, a charger with no working temperature compensation, and a room that does not move air. Ventilation and float control are the defenses.
How long do UPS batteries last?
VRLA UPS batteries commonly last about 3 to 5 years in service, up to roughly 10 on premium pure-lead designs. Flooded VLA cells can run 15 to 20 years with watering and maintenance. Lithium-ion often reaches 10 to 15 years. Heat, depth of discharge, and float control all move the real number, so trend each string.
How often should UPS batteries be tested?
The IEEE 1188 and 450 practice is monthly checks of float voltage, temperature, and a visual, quarterly ohmic measurements on every cell, and an annual capacity test, with capacity tested more often once a string ages or drops over 10 percent. Lithium relies on continuous BMS monitoring plus a yearly visual. The manufacturer and spec set the actual intervals.
Do lithium-ion UPS batteries need maintenance?
Lithium-ion UPS batteries need far less hands-on maintenance because the battery management system continuously monitors and balances the cells, so there is no watering, no equalize charge, and no per-cell float to set. The work becomes a yearly visual of the racks, review of the BMS state-of-health data, and a capacity check where the spec or warranty requires one.
Why does a UPS battery fail without warning?
A battery on float reads its full voltage and draws a trickle of current right up to the point it cannot carry a load, so voltage gives no warning. The warning lives in the capacity trend and the impedance trend, which is why a commissioning baseline and periodic ohmic readings matter. Without them, the outage becomes the test.
People also ask
Codes cited in this guide
This guide is written and reviewed against the published standards below. Always confirm the current adopted edition with the authority having jurisdiction.