I sat through a vendor demo a few years ago where the sales engineer spent forty minutes on a dashboard. Lovely dashboard. Live temperatures from every cabinet, colour-coded, updating every thirty seconds. At the end the operations manager asked one question: when this thing tells me a cabinet is drifting at eleven at night, who goes out? Nobody in the room had an answer. They bought the system anyway, and eighteen months later it was still emailing alerts to somebody who had left the company.
That is the honest starting point for smart commercial refrigeration. The technology works. The sensors are cheap, accurate and reliable in a way they simply were not ten years ago. What most operators get wrong is treating it as a technology purchase when it is really a process purchase, and buying it for the reason with the weakest evidence behind it.
What a smart refrigeration system actually is
Strip out the vendor language and there are four components. Sensors on the things that matter, usually air-on and air-off temperature, door state, compressor run time and sometimes discharge pressure. A gateway that gets that data off site. Somewhere to store it. And a rule set that decides what is worth waking someone up for.
The fourth part is where systems succeed or fail, and it is the part nobody demos. A cold room that swings two degrees during a delivery is behaving normally. A cold room that swings two degrees at four in the morning has a defrost problem or a door seal problem. Same data, completely different meaning. If the rule set cannot tell those apart, your team learns within a fortnight to ignore the alerts, and you have bought an expensive thermometer.
The part of this that is law, not marketing
Here is the argument I rarely see made properly, and it is the strongest one. For larger systems, automatic leak detection is not an upgrade. It is a legal requirement.
Under the GB F-Gas rules, leak checking obligations are set by the CO₂ equivalent of the charge in each refrigerant circuit, not the kilogram figure on the nameplate. You multiply the charge weight by the refrigerant's global warming potential and divide by a thousand. Government guidance on checking F gas equipment for leaks sets the frequencies, and states plainly that "you must fit a leak detection system if your equipment contains F gas equivalent to 500 tonnes or more of carbon dioxide".
| Charge in the circuit | Minimum leak check frequency | Automatic leak detection |
|---|---|---|
| Under 5 tonnes CO₂e | No routine requirement | Not required |
| 5 to under 50 tonnes CO₂e | At least every 12 months | Not required |
| 50 to under 500 tonnes CO₂e | At least every 6 months | Not required |
| 500 tonnes CO₂e and above | At least every 3 months | Mandatory |
Estates people are often surprised by how little refrigerant it takes to cross the first threshold, because the high-GWP legacy gases get there fast. That is the same pressure driving the R410A phase-out, and it is worth reading alongside the F-Gas rules for businesses before you scope any monitoring project. If you run a large pack system, part of your smart refrigeration business case has already been written for you by legislation.
What continuous monitoring does to your temperature records
Ask most operators how cold their chilled storage runs and they will give you a number straight away. Ask them to prove it for a specific Tuesday four months ago and the room goes quiet. That gap, between knowing and being able to show, is what monitoring actually closes.
The legal position is not vague. Regulation 4 of the Food Safety (Temperature Control) Regulations 1995 states that "no person shall keep any food...at or in food premises at a temperature above 8°C" where that food needs temperature control for safety. Most operators run cabinets colder than that deliberately, to leave headroom for door openings and defrost.
What changes with continuous logging is not the temperature. It is your ability to prove it. A manual check twice a day produces roughly sixty data points a month per unit and tells you nothing about the fourteen hours nobody was looking. A sensor logging every few minutes produces a continuous record, which means that when something does go wrong you can show exactly how long the excursion lasted and what was affected. That is the difference between discarding one delivery and discarding a room. We have gone deeper on this in our guide to cold room temperature logging, and on the legal temperature limits that sit behind it.
The energy claim, and where the evidence is thinner than the brochure
Almost every vendor leads with energy savings. I would treat that as the weakest part of the case, not the strongest, and I would rather tell you that now than have you discover it after the invoice.
A peer-reviewed study in Sensors built exactly this comparison, running a conventional cold store against an IoT-controlled one over three months. On product quality the modified system won clearly, with weight loss of "1.35 ± 0.09%" against "5.77 ± 0.17%" in the traditional store at one month, and materially better moisture retention at three. On energy, the same study reported no significant difference in electrical consumption between the two.
Read that carefully, because it is the useful finding. Monitoring on its own does not save energy. It tells you where energy is being wasted. The saving arrives later, when somebody acts on what the data shows: a door heater running in July, a defrost cycle firing six times a day when three would do, a condenser nobody has cleaned since installation. If your business case assumes the sensors themselves cut the bill, the case is wrong. If it assumes the sensors will find the faults that a maintenance team then fixes, it holds up.
Predictive maintenance, and what the data cannot tell you
Predictive maintenance is real, and it is oversold in roughly equal measure. Compressor run time creeping up week on week at a constant ambient is genuine early warning. So is a slow rise in the temperature differential across an evaporator, which usually means ice building where it should not. Those are patterns a human will not spot on a clipboard and an algorithm will spot in a fortnight.
What the data will not tell you is why. A rising superheat reading narrows the field to a handful of causes, and distinguishing a failing TEV from a partial restriction from an undercharge still means an engineer with gauges standing in front of the machine. Anyone selling you a system that diagnoses faults without a site visit is selling you a probability, not a diagnosis. Used properly the data changes what the engineer does when they arrive, because they turn up already knowing which circuit to look at. That is worth a great deal. It is not the same as removing the visit.
The same logic applies across a portfolio. On a single site, monitoring is a convenience. Across twenty sites it becomes the only sensible way to decide which one to attend first, which is why it tends to pay back fastest for people running cold storage maintenance at scale.
Buying this across multiple sites
If you are specifying monitoring across an estate rather than a single unit, a handful of decisions matter far more than the sensor spec, and they are the ones tenders usually leave out.
- Who receives an alert, and what happens if they do not act. Write the escalation path into the contract, not the handover pack. An unowned alert is worse than no alert, because it creates a record showing you knew.
- Whether the data is yours. Ask directly what happens to three years of temperature history if you change provider. If you cannot export it, you do not own your own compliance record.
- How it behaves when the internet drops. A gateway that buffers locally and backfills is worth paying for. One that simply loses the period is a gap in the exact record you bought it to produce.
- Whether it fits the plant you actually have. Mixed estates are normal. Cabinets of different ages from different manufacturers usually need a sensor-level approach rather than a controller integration, and vendors quoting on a clean modern site will not price for that.
- Who calibrates the sensors, and how often. An uncalibrated probe produces confident, official-looking, wrong numbers. That is worse than a clipboard.
Most of these end up defined in the servicing arrangement rather than the hardware purchase, which is why it is worth settling them while you are scoping a refrigeration maintenance plan rather than afterwards.
The refrigerant question underneath all of it
There is a tendency to treat the technology conversation and the sustainability conversation as separate. They are the same conversation, because the F-Gas phase-down is what forces the equipment decision.
Natural refrigerants change the arithmetic rather than just the environmental profile. CO₂ carries a GWP of 1, which means a system charged with it sits far below the thresholds that trigger quarterly leak checks and mandatory detection. Hydrocarbons behave similarly on paper, with flammability constraints that shape where they can be used. Against a legacy HFC charge, the compliance burden alone is a real operating cost you stop paying. The trade-off is that transcritical CO₂ systems run at much higher pressures and want engineers who have been trained on them, which is a genuine consideration in a market where that skill is not evenly spread.
Where this is actually heading
I am wary of predictions, so here are the three shifts I would bet on, all of which are already visible rather than speculative.
Monitoring stops being a product and becomes a default. It is already close to that in food retail, and it will move down into hospitality groups as the sensor cost keeps falling and insurers start asking what records you hold. Second, the compliance case overtakes the energy case in how these systems are sold, because the compliance case is provable and the energy case depends on somebody acting. Third, the interesting data stops being temperature and starts being everything around it: door discipline, delivery timing, how long a room takes to pull down after a big load. That is where the operational money is, and almost nobody is looking at it yet.
None of that requires new science. It requires operators to treat refrigeration data as something they act on rather than something they archive.
Frequently asked questions
Does a smart refrigeration system replace F-Gas leak checks?
No. Automatic leak detection changes the required frequency of manual checks for systems above the thresholds, and it is mandatory at 500 tonnes CO₂e and above, but a certified engineer still has to carry out and record the checks. Treat detection as something that works alongside the statutory regime, not instead of it.
Will monitoring actually cut our energy bill?
Only indirectly. The published evidence shows monitoring alone produces no significant change in consumption. Savings come from acting on what it reveals, so the business case should be built on the faults you expect to find and fix, not on the sensors themselves.
Can this be retrofitted to older cabinets and cold rooms?
Usually yes. Older plant often will not talk to a modern controller, so the practical route is independent sensors reporting to their own gateway rather than an integration with the existing controls. It is less elegant and it works on mixed-age estates, which is what most operators actually have.
How often do the sensors need calibrating?
Follow the manufacturer's schedule and record it, because a temperature record is only worth what the probe behind it is worth. This is the detail most often skipped, and it is the first thing a serious auditor will ask about.
What is the difference between temperature logging and a monitoring system?
A logger records and stores. A monitoring system records, evaluates against rules and escalates to a named person. If nobody is on the end of it, you have bought a logger regardless of what the brochure says.
Getting the specification right
The operators who get value from this are the ones who decided what they wanted to prove before they chose what to buy. If you are reviewing refrigeration monitoring across more than one site, it is worth talking through the plant you have and the records you need before the specification is written, because retrofitting a decision at that stage is expensive. Our team works on commercial refrigeration across London and the South East and is happy to go through it with you.
About the author: Ali Elm runs Be Cool Refrigeration & Air Conditioning, a family-run London firm that has been installing, servicing and repairing commercial refrigeration and air conditioning since 2004. His F-Gas certified team has delivered more than 2,000 projects across London and the South East.
Last updated: 20 August 2026
Written by
Ali Elm
Ali is the Head of Operations at Be Cool Refrigeration with over a decade of hands-on experience in HVAC and commercial refrigeration. He oversees every installation, repair, and maintenance project, making sure the work meets the highest standards. Ali holds full F-Gas certification and has worked across residential, commercial, and industrial refrigeration systems throughout London and the South East. When he is not on site, he writes these guides to help business owners and homeowners understand their cooling systems better.