Hydrogen trend in a battery room — what happens as a cell degrades

On the monthly round through a UPS battery room, a technician opens the door, holds a handheld gas meter up for a few seconds, glances over the terminal connections, signs the log and moves on. The reading sits in the normal range. The next round is a month away.

Nothing about that round is wrong. The problem is that for the month in between, there is no record of how any one cell behaved.

Why hydrogen is already there

A lead-acid cell under charge splits some of the water in its electrolyte, releasing hydrogen at the negative plate. That is not a sign of fault — it is a normal by-product of the charging chemistry, and battery-room ventilation is already sized around it.

So the real question is never “is there hydrogen.” It is level and trend over time. A cell going into overcharge, or one that is beginning to degrade, produces a measurable rise on that same channel — but it is a signal that is hard to pick out from the noise of a normal charge cycle with a single reading.

What a monthly round cannot show

Manual periodic measurement is not the wrong method. Its limit is in timing and frequency.

• A reading depends on where the cycle happens to be. Off-gassing rate varies with where a battery sits in its charge cycle; if the round lands on a low point, a rising trend is missed entirely.
• A single reading is not a trend. The difference between this month and last month says nothing about what happened on the days in between — it only shows the gap between two snapshots.
• Early on, the deviation is small. A cell that has only just started to degrade reads close enough to normal variation that it blends in.

Even when all three land right, there is still no record of what happened on the days between rounds.

What can be measured in between

AirQuality Sense (AQS-010) watches the hydrogen and TVOC channels continuously; the same probe reads temperature (±0.2 °C), relative humidity and pressure. Every channel passes through a Kalman filter and is reported both as an instantaneous value and as a trend — trend data stable enough to be used in occupational health and safety reporting.

A cell going into overcharge, or one beginning to degrade, produces a measurable rise on those two channels. The probe catches that change before it matures into a larger event.

Why a fixed threshold is the wrong instrument

If hydrogen were a gas that should never be present, a simple threshold would work. It is not — some amount is expected on every charge cycle. A system watching an absolute level either false-alarms on every cycle, or, once raised enough to stop doing that, also misses a real deviation.

Across the Cortex platform, each monitored point is evaluated against its own reference trend, learned during commissioning, rather than a fixed threshold. Applied to a battery room, that means the system is not asking whether hydrogen is present, but how far the current reading sits from that room’s own level during a normal charge cycle.

In short

A monthly round is a method that looks once a month, and it verifies the right thing. A cell’s degradation trend forms in the days in between. The two are not rivals: one gives periodic verification, the other continuity.

With a small team covering the site, the absence of the second means a problem is only noticed at the next scheduled round.