---
title: "Sensor calibration and maintenance: ensuring accuracy"
source_url: https://legionella.io/articles/sensor-calibration-and-maintenance-ensuring-accuracy/
canonical_url: https://legionella.io/articles/sensor-calibration-and-maintenance-ensuring-accuracy/
pillar: "Technology & Remote Monitoring"
summary: "A drifting Legionella temperature sensor can hide a failing outlet behind a green dashboard. How to catch silent drift and keep readings you can trust."
primary_keyword: "sensor calibration"
date_published: 2025-11-08
date_reviewed: 2026-06-26
author: "Legionella.io editorial team (REMOTE TECH LTD)"
reviewed_against: "HSE L8 and HSG274 guidance"
region: "United Kingdom"
license: "(c) REMOTE TECH LTD. Quote freely with attribution and a link to source_url."
---

# Sensor calibration and maintenance: ensuring accuracy

A sensor that has stopped working tells you something is wrong. A sensor that has quietly drifted tells you nothing useful — it keeps reporting a comfortable number that no longer matches the water. That second failure is the one that catches teams out, because the dashboard stays green while the control behind it slips.

Remote temperature monitoring only earns its place if the readings are true. Sensor calibration and maintenance are what turn a reading into evidence rather than a guess. Skip them, and you have spent money to be confidently wrong.

## Every reading is a compliance decision you've already made

When a sentinel tap, a calorifier flow-and-return pair or a cold-water storage tank is wired to a sensor, each number it sends becomes a small decision. "Hot water reached the outlet above the expected temperature, so no action." "Cold stayed under the expected ceiling, so no action." Those decisions are only as good as the instrument making them.

HSE guidance expects hot water to be stored hot and distributed so it reaches outlets quickly, and cold water to stay genuinely cold — commonly framed as storage around 60°C, distribution reaching about 50°C within a minute, and cold below roughly 20°C [1]. A sensor reading two or three degrees high turns a failing outlet into a passing one on paper. Nobody flushes it, nobody investigates, and the record says everything was fine. Temperature sensor accuracy is therefore not a maintenance afterthought; it is the difference between a logbook that proves control and one that quietly launders a problem.

## A green dashboard that wasn't (an illustrative scenario)

The following is a composite, drawn from a pattern that recurs rather than a single named site.

A mid-sized office retrofit added remote temperature monitoring across a couple of dozen sentinel outlets, plus the calorifier and the cold tank. For the first year the monthly review was uneventful. Hot outlets reported in the low 50s, the calorifier flow sat just above 60°C, exceptions were rare and quickly closed. The responsible person reasonably treated the system as under control.

What the dashboard never showed was that one sensor on a far hot outlet had drifted. Over many months its readings crept upward relative to reality — a slow offset, not a sudden jump — so no alert ever fired. The sensor reported 51°C; a hand-held check at the tap, done much later, found the water nearer 45°C. That outlet had been sitting in the growth-supporting band the whole time, logged every month as compliant.

The drift surfaced almost by accident. A contractor doing planned servicing took a spot reading with a separate thermometer and noticed the gap. One outlet became a question; the question became a recalibration of the whole estate; the recalibration found a second sensor with a smaller offset in the same direction.

Nothing dramatic happened to anyone. But the records for that outlet had been wrong for the better part of a year, and the team could no longer say with confidence when control had actually held. That is the real cost — not only the drift, but the loss of trust in the history.

## What that scenario was really about

Three decisions, taken earlier, would have caught it.

The first is verification against an independent reference. A monitoring sensor checked only against itself can drift indefinitely without contradiction. A periodic comparison reading — a hand-held thermometer at the same point, that thermometer itself calibrated against something traceable — gives the sensor something to be wrong against. When the two disagree beyond a sensible tolerance, you have found the drift before it buries itself in the record.

The second is treating slow offset as a genuine failure mode, not only sudden dropout. Most alerting is built to flag a sensor that goes silent or reports obvious nonsense. Gradual sensor drift produces neither. It needs a scheduled calibration check, not just an alarm threshold.

The third is recording the calibration decision, not just the event. "Sensor X checked, within tolerance" is weaker than "Sensor X compared to reference at the outlet, read 0.4°C high, inside our 1°C tolerance, next check due [date]." The second version lets the next person see the trend forming.

## Keeping sensors honest, in practice

You do not need a metrology lab. You need a short, repeatable routine folded into the existing scheme of control.

- **Set a calibration interval and write it down.** Start from the manufacturer's stated interval and record it in your written scheme; sensors in hot, scaling or hard-water conditions may warrant checking sooner.
- **Verify against an independent, calibrated reference.** Compare the live reading to a hand-held thermometer at the sensor's point of measurement, and log both figures and the difference — not simply a pass.
- **Watch direction, not just size.** A small offset that grows month on month matters more than a one-off blip. Trend the comparison readings over time.
- **Maintain the physical install.** Scale, a loosened probe, a sensor reading pipe surface instead of water, or a unit relocated during other works can all skew results. Check placement when you check calibration [2].
- **Close the loop on out-of-tolerance sensors.** Decide in advance what happens when a sensor fails its check: recalibrate or replace, then re-examine the readings it produced since its last good check.

Done this way, the data feeding your monthly reviews and any [remote compliance audit](https://legionella.io/articles/remote-audits-digital-compliance-checks-from-afar/) carries weight, because you can show why you trust it. If you are still deciding whether to bring this kind of Legionella temperature monitoring in at all, [the guide on upgrading your programme](https://legionella.io/articles/upgrading-your-programme-when-to-adopt-new-technology/) covers when the technology is worth adopting in the first place.

## Where guidance ends and your judgement begins

The temperatures above are general HSE expectations, not a replacement for the figures your own risk assessment sets, and the tolerances and intervals you adopt are yours to justify through a competent, site-specific assessment [3]. A sensor reading — even a freshly calibrated one — is evidence to act on, not proof on its own; it sits alongside flushing, cleaning, inspection and, where appropriate, sampling. Calibration is also governed by what the instrument's manufacturer specifies for that device, which outranks any rule of thumb. If you are unsure whether your monitoring data would stand up to scrutiny, treat that uncertainty as the finding to act on.

## FAQ

### What is the difference between a sensor failing and a sensor drifting?
A failed sensor usually stops reporting or sends an obviously wrong value, so it raises an alert and gets fixed. A drifting sensor keeps reporting plausible numbers that are gradually wrong. Drift is the more dangerous of the two precisely because it does not announce itself, which is why scheduled calibration checks matter more than alarms alone.

### How often should temperature sensors be calibrated?
Begin with the manufacturer's stated interval and record it in your written scheme, then tighten it for sensors in hot, scaling or otherwise harsh conditions. Your risk assessment, rather than a fixed universal figure, sets what is reasonable for your system [2][3].

### Does remote monitoring replace manual temperature checks?
No. Independent hand-held readings are what let you verify the sensors in the first place. Remote monitoring reduces routine manual effort and surfaces problems faster, but it depends on periodic manual comparison to stay trustworthy [2].

## Related reading

- [Upgrading your programme: when to adopt new technology](https://legionella.io/articles/upgrading-your-programme-when-to-adopt-new-technology/)
- [Remote audits: digital compliance checks from afar](https://legionella.io/articles/remote-audits-digital-compliance-checks-from-afar/)
- [Future-proof design: plumbing designs that minimise Legionella](https://legionella.io/articles/future-proof-design-plumbing-designs-that-minimise-legionella/)

## Sources

[1] HSE, "Hot and cold water systems". https://www.hse.gov.uk/legionnaires/hot-and-cold.htm
[2] HSE, "Legionnaires' disease: Technical guidance (HSG274)". https://www.hse.gov.uk/pubns/books/hsg274.htm
[3] HSE, "Legionnaires' disease. The control of legionella bacteria in water systems - Approved Code of Practice and guidance (L8)". https://www.hse.gov.uk/pubns/books/l8.htm