UPS and Battery Monitoring: What to Measure Unattended, How to Tier Alarms

Five risks that show up when UPS and batteries run unattended

• Walk-around checks miss early internal-resistance fade: a cell that still looks sound may already have climbed in resistance, and by the time terminal voltage has clearly dropped, backup time has usually already shortened
• One lagging cell pulls the string down: a single open or weak cell in series cuts the string’s ability to discharge, and the real backup gap appears only at the next outage
• The thermal-runaway window is short: from local heating to smoke, the minutes available for action are few, and nobody is standing in an unattended room to see it
• Replacement is mistimed both ways: changing the whole string on calendar age writes off cells that still work; delaying change leaves the risk sitting until the next outage
• After the outage, the cause cannot be named: without cell-level history it is impossible to say which cell or which string dropped during discharge, and the review stops at “the string failed”

All five rest on the same condition: nobody is watching the instruments in the room. To move fade from “found after the fact” to “raised by a rule” in an unattended room, collection has to reach the cell, not only the string voltage on the UPS panel.

What cell-level monitoring has to collect

If “online battery monitoring” is only a UPS communications port showing string voltage and load percentage, it does little to locate a lagging cell. To decide which cell to replace and which string is still in a normal float band, collection at least needs cell voltage, internal resistance and temperature, aligned on the same time axis with string current and float status.

ORPAON can provide cell-level battery monitoring, three-tier alarms, a B/S platform, and energy-storage three-layer BMS with fire and power-cut linkage as solution capabilities; scope is confirmed per project. Whether the module matches the battery type, how the bus is laid and how thresholds are split are written into the technical solution only after a site check. A point list from another site should not be copied across. ISO 50001 work, where it applies, can reuse UPS-state series as part of the energy picture; it does not replace the monitoring design described here.

• Cell voltage: find a lagging cell, an open circuit or a poor connection; string voltage alone hides a single-cell deviation in the average
• Internal resistance: a commonly used early-fade indicator; appearance and terminal voltage can still look normal after resistance has already left the string reference
• Temperature: local heating is an observable signal before thermal runaway; read both a point near the cell and the room ambient
• String current and float status: a Hall sensor or other current pickup separates charge, discharge and float so voltage movement during those modes is not treated as a fault
• Consistency: spread of voltage and resistance inside the string shows whether the issue is one cell or end of string life, which avoids replacing the whole string without cause

Alarm tiers and fire / trip linkage (solution capability)

Collection without tiers turns alarms into noise the shift learns to ignore. A recommended approach is three tiers — watch, schedule a replacement, attend immediately — with the recipient and the channel written down in advance. Sounder, voice call, SMS and email can run together, but the responsibility chain has to end on a shift and a role, not only on a large screen.

Fire and power-cut linkage is a solution capability, not a field result that ships by default with the monitoring software. Smoke, heat and combustible-gas detection, plus very-early air sampling, and whether they interlock with the BMS, HVAC and a shunt-trip circuit, are confirmed in the project against electrical and fire rules and the owner’s procedure. Whether the control room can issue fire-related commands is likewise fixed by the interfaces and permissions both sides accept.

• Tier one: cell voltage or resistance leaves the string reference, room temperature approaches the agreed upper limit — send to the inspection role, add to the next on-site check
• Tier two: spread is widening, resistance keeps climbing, local heating is accelerating — send to the maintenance owner, book a shutdown window for replacement rather than waiting for the next outage
• Tier three: heating rate or gas detection hits the agreed rule, thermal-runaway or fire risk is present — attend immediately and start the fire and power-cut actions written in the procedure
• Linkage objects: smoke / heat / combustible gas, very-early air sampling, HVAC stop, shunt-trip disconnect; which circuits are brought in and who may confirm an action are written into the project interlock table
• View and push: a B/S platform for live and history; sounder, phone, SMS and email split by tier and channel; a push with no named owner is not counted as covered

Monitoring object, collection points and alarm use

Before a proposal is written, line up “what is collected” with “what the alarm is for”. All three columns should be checked on site against battery type, UPS communications and fire interfaces. The rows below are a working split, not a fixed delivery list.

Thresholds, scan period and interlock objects change with battery chemistry, room conditions and the owner’s procedure. Before reuse on the next site, match the monitoring module to the battery type again and review electrical, fire and BMS / PCS interfaces.

Where battery monitoring stops and energy-storage EMS begins

UPS and battery monitoring answers whether this cell can still discharge at the next outage and whether it will run into thermal runaway first. Energy-storage EMS answers when the plant charges and discharges and how SOC and the point-of-connection limits are held. Both sit on the safety chain; the deliverables differ. One is cell-level watch and tiered alarms; the other is energy dispatch and an equipment-domain model. Writing both needs onto one work order makes scope and acceptance criteria diverge.

The in-house EMS family is split by form: the on-site WPF HMI form has deployment evidence; the web management platform is an extensible form, assessed per project scope; the cross-platform client and large-language-model dispatch suggestions are in R&D and joint validation and are not a standard delivery commitment. Even when suggestions are in a joint-validation scope, a person on shift confirms them before they are executed; an automatic closed loop must not be described as a standard delivered function.

• Monitoring side: cell and room collection, three-tier alarms, B/S view and push; acceptance looks at the point list, thresholds, the responsibility chain and whether history can be traced
• Energy-storage EMS side: BMS / PCS / meter domain connection, charge/discharge strategy, SOC hold; the on-site WPF form has deployment evidence, the web form is assessed per project
• Not in standard delivery: the cross-platform client, LLM dispatch suggestions, and treating design figures as accepted field results
• Project entry: inventory UPS units, battery types, whether a BMS interface already exists, and whether fire and shunt-trip circuits may be connected — then decide which stretch monitoring covers and which stretch EMS covers

Design objects and design figures must not be written as field implementation, acceptance or financial results. Figures without site evidence are not published in this article; later projects must reconfirm battery type, the alarm responsibility chain and electrical / fire / PCS interfaces.