---
title: "Drones and robots for inspecting water systems"
source_url: https://legionella.io/articles/drones-and-robots-for-inspecting-water-systems/
canonical_url: https://legionella.io/articles/drones-and-robots-for-inspecting-water-systems/
pillar: "Technology & Remote Monitoring"
summary: "When a drone or tank-inspection robot earns its keep on a UK water system, what its footage actually proves, and where it is just expensive theatre."
primary_keyword: "drones water inspection"
date_published: 2025-12-08
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."
---

# Drones and robots for inspecting water systems

Somewhere on most large UK sites there is a tank nobody has looked inside for years. Not through negligence, but because looking inside means isolating the supply, draining a compartment, and sending a person into a covered space that counts as a confined space the moment they climb in. So the internal inspection the risk assessment quietly asks for slips from "annual" to "when we get round to it". That gap is exactly where a tank-inspection robot or a survey drone changes the maths.

This is not about gadgets. It is about access. A drone or a remotely operated camera earns its place when the thing you need to inspect is high, enclosed, or hazardous to reach, and it stays parked in the cupboard the rest of the time.

## A tank nobody wanted to open

Picture a college campus with a sectional cold water storage tank in a rooftop plant room, split into two compartments so one side can be isolated for maintenance. (This is a composite scenario, drawn from common arrangements rather than a single named site.) The internal inspection had been deferred three years running. Each time it surfaced, the cost of draining a compartment, arranging confined-space entry and taking part of the building off mains water pushed it back down the list.

A new responsible person took a different route: a tethered inspection camera lowered through the existing access hatch, with a small crawler head to move across the tank floor. No draining. No entry. Power and lighting ran down the same umbilical, with the footage recorded at the surface.

What it showed mattered. One compartment had a thin layer of sediment across the base and an ill-fitting lid section letting daylight and dust in, enough to explain a slow warming and the grit that kept turning up at a nearby outlet. The second compartment was clean. That single distinction changed the job from "drain and clean the whole tank" to "clean one compartment, repair one lid, leave the rest in service".

## The decisions that actually mattered

The robot did not fix anything. The decisions around it did.

The responsible person wrote down *why* the inspection was done by camera, namely deferred access, confined-space risk and supply continuity, and tagged the footage to the specific asset and date. That turned a video clip into an inspection record of the kind L8 expects duty holders to keep alongside their other monitoring evidence [1]. A clip with no asset reference and no date is just a clip.

They disinfected the camera and crawler head before it went into stored drinking water. A dirty inspection tool in a wet tank is a contamination route, not a control. The kit has to come out cleaner than it went in, not the other way round.

And they resisted the obvious trap. Clear footage of a clean-looking compartment is not a clean bill of health. The temperature regime, the flushing of low-use outlets and the sampling the risk assessment called for all carried on regardless; HSE is explicit that testing follows the system and the risk assessment, not the reassurance of a tidy-looking tank [4]. The footage justified a lid repair and a targeted clean. It did not retire a single control.

## Where this kind of kit genuinely helps

Strip away the novelty and the honest use cases are narrow but real.

- **High or roof-mounted plant.** Cooling towers and roof tanks are both awkward and high-risk; cooling towers in particular sit among the systems HSE flags as most likely to create Legionella risk [3]. A drone can survey pack condition, drift eliminators, fan housings and general fabric from the air far faster than rigging access, helping you decide whether a closer hands-on inspection is warranted.
- **Large covered storage tanks.** A tethered camera or crawler lets you judge internal condition, including sediment, corrosion, ingress, and lid and screen integrity, without draining or confined-space entry, so the inspection actually happens on schedule instead of slipping.
- **Building elevations and discharge points.** A drone pass can spot an overflow or warning pipe that has been running, or insulation that has failed on an external run, without a cherry picker.

Notice the pattern. In every case the binding constraint is access, whether height, enclosure, or risk to a person, not the inspection itself. Where you can simply lift a hatch and look, a robot is theatre.

## What it does not do

It is just as important to name the limits. A drone does not read a temperature at an outlet. A crawler does not flush a dead leg or clean a calorifier. None of this equipment tells you whether Legionella is present, because that is a laboratory question, and a clear shot of clean-looking water proves nothing about what is living in the biofilm on the wall. The routine tasks set out in HSG274, including temperature checks, flushing, tank and outlet inspection and cleaning, still belong to people [2]. The robot widens what you can see; it does not shorten the list of things you have to do.

There is a quieter risk too: letting impressive footage crowd out unglamorous control. A four-minute tank video gets attention in a management meeting. Weekly flushing of a rarely-used shower does not, and yet it is the second one that prevents cases. Use the imagery to direct effort, not to replace it. When an inspection does turn up a fault, the value is in closing it out properly, which is harder than it sounds; see [Failed remedial action close-out](https://legionella.io/articles/failed-remedial-action-close-out-why-positive-results-keep-returning/) for why findings have a habit of coming back when the fix is never verified.

## Before you spend on it

A drone or tank-inspection robot is an inspection aid, not a control measure. It can show you the inside of a tank you could not safely reach; it cannot keep that tank cold, flush a stagnant leg, or tell you whether the bacteria are there. Treat the footage as one input to a competent, site-specific risk assessment. The inspection interval, the acceptable condition, and the response to anything it reveals are decisions for your responsible person and your assessor, not the equipment supplier.

The next step is small. Pull your asset register and mark every water asset you currently inspect badly, rarely, or at real risk to a person: the roof tank, the cooling tower, the covered store you keep deferring. If that list has two or three entries, you have found where remote inspection might earn its keep, and you can put a sharp question to suppliers instead of a vague one: "show me how your kit inspects *this* asset, and how the output lands in my records." If the list is empty, your money is better spent elsewhere.

## FAQ

### Can drone or robot footage replace a physical tank inspection in our records?
It can stand in for the *access*, not the judgement. Footage counts as a genuine inspection record only when a competent person interprets it, it is dated and tied to the specific asset, and any fault is logged and acted on, which is the same record-keeping L8 expects of any inspection [1]. An unlabelled video sitting on someone's phone is not evidence of anything.

### Is it safe to put a robot inside a stored drinking-water tank?
Only if everything that enters the water is clean and disinfected first. Inspection gear that has been in other tanks, or stored without cleaning, can carry contamination straight into potable water. Agree the cleaning and disinfection method with whoever operates the kit before it goes anywhere near a wholesome-water tank, and record that it was done.

### Where do drones actually beat sending a person up?
Where access is the problem. Roof-mounted cooling towers, high-level tanks and external elevations are the clear wins: a drone surveys condition quickly and keeps people off ladders and out of confined spaces, then tells you where a closer hands-on inspection is genuinely needed [3]. For anything you can reach by opening a panel, the old-fashioned look is cheaper and just as good.

## Related reading

- [Failed remedial action close-out: why positive results keep returning](https://legionella.io/articles/failed-remedial-action-close-out-why-positive-results-keep-returning/)
- [Emerging biosensors for Legionella detection](https://legionella.io/articles/emerging-biosensors-for-legionella-detection/)
- [Machine learning for Legionella risk prediction](https://legionella.io/articles/machine-learning-for-legionella-risk-prediction/)

## 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, "Systems most likely to create legionella risk". https://www.hse.gov.uk/legionnaires/risk-systems.htm
[4] HSE, "Testing and monitoring your water system for legionella". https://www.hse.gov.uk/legionnaires/testing-monitoring-water-system.htm