A clipboard check tells you one thing: a tap read 53°C at 9.40 on the morning someone walked round with a probe. It says nothing about the other 700-odd hours that month. Remote temperature monitoring closes that gap — not by handing you more single readings, but by showing whether the system actually held temperature when nobody was looking.
That is the real benefit, and it is also where teams spend on the wrong thing. Continuous data is only worth fitting if it changes a decision. A sensor with no threshold and no owner is just a more expensive clipboard. So the useful question is not “should we go digital?” but “which points, which alerts, and who acts when one fires?”
What continuous data actually buys you
The honest list of benefits is shorter than the brochures suggest, but each item is real.
It catches drift between checks. Manual monitoring samples a moment; water temperature sensors watch the gaps, so you see the cold-water creep on a warm bank-holiday weekend or the calorifier that dips under peak morning draw-off. It cuts legwork. Walking a probe round a large estate is hours of skilled time; automated temperature monitoring frees most of that for the parts of the job a person genuinely has to do. And it builds a continuous evidence trail, which matters when you have to show a controlled system rather than just assert one.
What it does not do is run your scheme for you. The kit reports temperature; it does not decide whether a reading is safe, clear a dead leg, or descale a shower head. Keep that line clear and the rest of the setup falls into place.
What you need in place first
Sensors enforce a control scheme. They do not create one. Before any kit goes on a pipe, you want three things settled.
A current risk assessment and written scheme that already names your control points and target temperatures — the monitoring mirrors that, it does not invent it. A clear list of where temperature actually matters on your system. And a named person who will read the alerts and own the response. Skip the third and you will have a beautiful dashboard nobody is accountable for.
A setup sequence that holds up
Treat the rollout as ordered steps, each finished only when it meets a clear test. The “done when” line is the point — it stops a half-finished install passing for a working one.
-
Map the points to the scheme. List the control points your written scheme already relies on: calorifier flow and return, the cold storage tank, the sentinel outlets nearest and furthest from each source, and the low-use or awkward-to-reach outlets you keep forgetting by hand. HSE technical guidance frames monitoring around representative points like these rather than every tap [2]. Done when every proposed sensor maps to a specific line in the written scheme.
-
Match the sensor to the point, and confirm the data can leave the plant room. A clamp on a hot return and an in-line probe at a blended outlet answer different questions, so choose per point — surface vs in-line, battery vs mains, and the network it reports over. Check each one reads the water temperature you care about, not pipe metal or ambient air. (See IoT sensors for water system monitoring: an overview for how the sensor types differ.) Done when every point has coverage and a tested path for its data out of the building.
-
Set thresholds and escalation to your scheme, not the device defaults. Configure alarms to the targets and tolerances in your scheme, then route each temperature alert to a named person with a defined action — not a shared inbox. Done when no alert exists without an owner and a specific “do this next” response.
-
Baseline against a known-good probe. Run the sensors alongside a calibrated handheld for a period and reconcile any difference. Done when remote and handheld readings agree within an acceptable margin at the same point and time.
-
Feed it into the record the duty holder relies on. Connect the feed so readings, exceptions and the actions taken all land in one auditable place rather than a separate app nobody opens. A central Legionella risk-management platform is where this usually lives. Done when an inspector could trace an out-of-range reading to the action taken without asking you.
-
Run parallel before retiring manual checks. Keep manual monitoring going until the system has proven itself over a meaningful stretch, and keep a fallback for the day a sensor or its comms drop out. Done when you trust the data and have a written plan for what happens when a sensor goes dark.
Where these setups quietly fail
Most disappointments trace back to one of a handful of mistakes, and none of them are about the hardware being bad.
The commonest is location. A sensor reading pipe surface, trapped air or a dead point gives a confident number that means nothing. Next is alert fatigue: set too many thresholds too tight and people start ignoring the lot, including the one that mattered. Then there is dashboard-as-control — graphs everyone admires and nobody is responsible for. And finally, drift and downtime: batteries flatten, calibration wanders, and connectivity drops, so the monitoring itself needs monitoring. Plan a check on the kit, not just the water.
Proving it is actually working
Verification is the step people skip, and it is what separates real Legionella temperature control from expensive decoration. Reconcile against a calibrated probe on a periodic basis. Audit the alert-to-action trail: pick a recent exception and follow it to who was told and what they did. And look for the patterns only continuous data exposes — the void room that never gets flushed, the far wing that loses its hot water by mid-afternoon, the weekend the cold tank warmed past target. If the system is surfacing those and someone is acting on them, it is doing its job.
What the technology does and does not settle
None of this replaces a competent person or a live risk assessment. A sensor states a temperature; it does not judge whether that temperature is safe, repair a stagnant leg, or clean a tank. Calibration drifts, batteries die and links drop, so build verification into the monitoring as well as the water. Remote data counts as evidence of control only when someone competent reads it, acts on the exceptions, and records both. The kit sharpens that judgement — it is not a substitute for it. Where the data points to a wider problem, that is a trigger to revisit the assessment, not just silence an alarm [1].
FAQ
Does remote monitoring let us stop manual temperature checks?
Not immediately, and not entirely. Run both in parallel until the automated system has proven reliable, then your risk assessment decides what manual work remains. Calibration, physical inspection and tank condition checks still need a person on site [1].
Where should the first sensors go?
Start with the points your scheme already leans on: calorifier flow and return, the cold storage tank, and the sentinel outlets nearest and furthest from each source. Add the low-use outlets you most often forget to check by hand, since those are where temperature problems hide [2].
What temperatures should the system flag?
Set alarms to the targets in your written scheme. As a general benchmark, HSE’s hot and cold water guidance points to cold water kept below 20°C where practicable, hot water stored at around 60°C, and roughly 50°C reached at the outlet within a minute of running — 55°C in healthcare premises. Confirm the exact figures for your system through the assessment [3].
Will the HSE accept remote data as proof of control?
What matters is whether the data shows a controlled system and a responsible person acting on exceptions; the format is secondary to the management behind it. Continuous readings with no owner or follow-up prove less than a tidy manual log someone clearly acts on [1].
Sources
[1] 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 [2] HSE, “Legionnaires’ disease: Technical guidance (HSG274)”. https://www.hse.gov.uk/pubns/books/hsg274.htm [3] HSE, “Hot and cold water systems”. https://www.hse.gov.uk/legionnaires/hot-and-cold.htm