2026-09-01

Telecom Battery Maintenance: From Scheduled Visits to Exception-Based

Battery ESS48V Telecom BatteryOperations

Scheduled battery inspections have a structural weakness: they only find problems on the day someone is standing at the site. Remote telemetry changes that — and for operators with spread-out sites, it usually changes the economics more than the battery chemistry does.

What scheduled visits actually cost

A periodic inspection regime has three costs that rarely appear in the battery budget. The direct cost of the visit — travel, labour, access. The blind spot between visits, where a fault develops and goes unnoticed until the next scheduled check or until an outage exposes it. And the wasted visits to batteries that are perfectly healthy, which is most of them.

The alternative is not "no maintenance". It is exception-based maintenance: the battery reports continuously, and crews are dispatched when something is actually wrong.

The data you need for it to work

Battery data required for exception-based maintenance
Data point	What it tells you	How it is used
State of charge	Energy available right now	Runtime estimates, low-charge alarms
State of health	Capacity remaining versus new	Replacement planning and budgeting
Cell voltages	Balance across the pack	Early warning of a failing cell
Cell and ambient temperature	Thermal stress	Derating, life estimation, cooling faults
Alarm states	Protection events	Dispatch trigger

All five come from the BMS. The question is only how they travel — over RS485 to a local controller, or over SNMP straight into the NMS.

Designing the alarm set

More alarms is not better. A battery that raises forty trap types into a queue nobody reads is not monitored, it is noisy. Start with the five that change what you do:

Low SOC — set above the shutdown point, so there is still time to act
Cell voltage deviation — imbalance appears long before capacity loss does
Temperature out of range — both high (life) and low (deliverable capacity)
SOH below threshold — the trigger for planning replacement
Communication loss — so silence is not mistaken for health

From alarms to a response

An alarm is only useful if it has an owner and a response. For each of the five above, decide who receives it, what they do, and what the target response time is. Without that, the telemetry becomes a dashboard nobody acts on.

Planning replacement on trends, not on failure

State of health is the number that makes replacement predictable. Instead of replacing strings reactively — usually after an outage reveals a weak battery — you watch the SOH trend across the estate and replace the sites that are approaching the threshold, in a planned programme, with budget allocated in advance.

That is a genuinely different operating model: battery replacement becomes scheduled capital work rather than emergency call-outs.

What it looks like in practice

At three telecom sites in Pakistan, 48V 100Ah packs report over SNMP into the operator's NMS. Health is checked from the operations centre rather than on site, and because temperature is reported continuously, the effect of sustained heat on the packs is observed rather than estimated — which matters when the customer's real concern is service life.

Note that the saving here is not only the avoided visits. It is also that the sites which do need attention are identified while the problem is still small.

Getting there in four steps

Specify BMS telemetry as a requirement on the next battery purchase — it is far cheaper than retrofitting.
Decide the alarm set and the response owner for each alarm.
Run telemetry alongside existing scheduled visits for one cycle, to build confidence in the data.
Extend the interval between routine visits, keeping them for the physical checks telemetry cannot do.

More on the technical setup is in the SNMP monitoring guide; the commercial comparison against lead-acid is in the total cost math.

FAQ

Does remote monitoring remove the need for site visits entirely?

No. It removes the routine health-check visits. Physical inspection — terminations, mounting, ventilation, cleaning — still needs a person, but it can run on a much longer interval.

What if the network itself is down?

Then the battery is not reporting, which is what the communication-loss alarm is for. Design the system so that loss of telemetry is itself an event, rather than an absence that goes unnoticed.

How accurate is state of health?

It is an estimate derived from measured capacity, impedance and cycling history, and it is good enough for trend-based planning — which is the use that matters. Treat it as a trend indicator rather than a laboratory measurement.

Can this work on an existing VRLA estate?

Partly. External monitoring can track voltage and temperature of a VRLA string, but cell-level data and true SOH are not available the way they are from a lithium BMS. Most operators introduce exception-based maintenance alongside the move to lithium.