Battery String at End of Life: Patch It or Replace It
Two 1.0 MW UPS systems each support two strings of 40 twelve-volt VRLA blocks, 160 blocks in total, commissioned in 2019 against a five-year design life at 25°C. The battery room runs 28–30°C through summer. The annual conductance survey shows nine blocks below 80% of baseline, three below 70%, and one block at 45% sitting in a string whose other blocks test near baseline. Last year's load-bank test gave 8 minutes of runtime at full load against a 10-minute design. Vendor A quotes total replacement at €340 per block (€54,400) plus disposal; Vendor B proposes replacing the twelve failing blocks for €4,100. The maintenance contract renews in six weeks and the capex cycle closes in three months. The facilities director wants the cheaper option and asks you to confirm the batteries will hold until the next budget year.
1. Evaluate the two proposals using battery end-of-life criteria and the operating environment, and recommend a course of action with justification.
Read the worked answer
Start from the acceptance criteria. Industry practice (IEEE 1188 for VRLA) treats a block at or below 80% of its baseline conductance, or a string whose measured capacity has fallen below 80% of rated, as failing. Here three blocks are below 70% and one at 45% - well past that line. The load-bank result is the stronger evidence: 8 minutes against a 10-minute design is 80% of rated capacity at the boundary, and load-bank results drift downward, not upward.
Correct the age calculation. Five years is quoted at 25°C. At 28–30°C the ageing rate roughly doubles for each 10°C above reference, so the array has accumulated the equivalent of six to seven years at reference temperature. Effective service life is already exhausted.
Why partial replacement is dangerous. Mixing new blocks into an aged string creates unequal float voltages: the new blocks run at a different internal resistance, so the charger either overcharges them or undercharges the aged blocks, and the new blocks fail early. Vendor B's proposal also leaves the string's weakest remaining blocks in place and leaves the temperature problem untouched, so it buys months, not years.
Recommendation. Do not certify the batteries as fit until the next budget year. Approve string-level replacement of the two worst-performing strings this quarter, with the remaining strings funded in the next cycle, and make the room cooling correction a condition of the warranty. Require a capacity test after replacement, not just conductance. Document the residual risk on the risk register with a runtime-derived exposure for the load actually carried.
2. What interim risk mitigations and maintenance-contract changes should be in place while the replacement is staged?
Read the worked answer
Reduce the consequence of failure rather than pretending the risk is gone. Quantify what the current 8 minutes actually covers: compare it with generator start and transfer time plus the load at the time. If a single generator start-to-transfer sequence is 15 seconds and the generator has a known failure mode, battery runtime is not the real protection - the generator is. So the interim package should combine:
- Load reduction or transfer of non-critical load off the affected UPS to extend runtime and lower discharge stress.
- Monitoring: string-level and block-level impedance/voltage monitoring with alarm thresholds, plus quarterly torque and connection checks - most VRLA failures in service are open-cell or connection failures, not gradual capacity loss.
- Test discipline: annual load-bank test with recorded per-string results, and a documented decision rule that a second consecutive degradation triggers mandatory replacement.
- Environmental correction: cooling set-point and airflow correction in the battery room, recorded so the warranty claim is defensible.
Contract changes at renewal. Add: a defined spares holding for the installed block type; response and restoration times specific to UPS and battery systems, not generic; a requirement for conductance surveys with the contractor's raw data, not a pass/fail summary; a temperature-excursion notification obligation; disposal and environmental compliance certification; and an escalation clause if the contractor's own survey recommends replacement and the client defers. Remove any clause that lets the contractor certify a string as healthy on float voltage alone.
Governance. Put the deferral decision in writing with the accountable owner named, so that if a string fails during a mains outage the organisation can show a documented, quantified acceptance rather than an omission.