Commercial refrigeration and extraction monitoring across a multi-site franchise

The refrigeration equipment across these kitchens already held temperature and, in most cases, already displayed it. What nobody had was real-time visibility of that temperature across multiple locations at once, and no way at all to see why a reading had moved. The concerns the consultancy raised were specific: temperature fluctuation caused by doors left open, and how long those doors were being left open.

The second of those is the harder one, and it is where most cold storage monitoring stops. A plain door contact reports open and closed. On that basis a door held open for 30 seconds while someone reaches for a tray and a door standing open for 30 minutes during a delivery produce the same record, because both of them are simply open. The difference between those two events is the whole question, and an open or closed state cannot express it.

The kitchen extraction side had a different shape of the same problem. Commercial filters in the extraction path were serviced to a schedule, and a schedule is set in advance of the thing it is meant to respond to. A filter serviced to a calendar is serviced early or late, and neither condition is visible from the calendar. Cooking load varies by site and by week, so the interval that suits one extraction system is wrong for the next one.

Both halves come back to the same gap. The equipment produces a signal at the point it sits, and the people responsible for it are somewhere else.

Every tag transmits and never receives. Readings travel one way, over 433 MHz radio, to a gateway at the location, and the gateway's cellular connection carries them on from there. There is no downlink to the tags, so nothing is sent back to them: no thresholds, no configuration and no commands. Threshold evaluation, alarm delay, escalation and acknowledgement are all server-side, which is where they can be changed without anyone going near a kitchen.

Reporting runs on two paths at once. A state change is transmitted immediately, and that is the path a door event or an alarm condition takes. Underneath it, each monitored point reports its status on a schedule. The scheduled cycle is the health path rather than the response path, and its value is that silence becomes legible: a report expected within the period that did not arrive is itself the alert.

Open duration is derived server-side from the state change, not measured at the door. The tag reports that the door opened and, later, that it closed; the platform holds both timestamps and the interval between them. This is the point on which the whole refrigeration requirement turns, and it is worth being precise about where the work happens. The door contact reports events. The duration is arithmetic done afterwards on the record of those events.

Placement is engineering rather than preference. The probe and the door contact sit inside the refrigerated space, because that is where the measurement is. The radio enclosure is positioned outside the insulated enclosure wherever it is practical to do so, because metal panels and foil-faced insulation attenuate, and a transmitter sealed inside a lined steel box has to work through the one material in the room designed to stop things passing through it.

On the extraction side the measurement is taken across the filter assembly rather than at it. Tapping points upstream and downstream of the filters run by tubing into the duct, and the reading is the difference between the two pressures. That difference is what changes as a filter loads, which is why it is the measured quantity rather than either pressure on its own.

Door events arrive with a duration attached. A 30-second opening and a 30-minute one are separate records rather than the same record seen twice, and that separation is what the refrigeration requirement was asking for. Everything else on this side of the engagement is built on it.

Duration also makes two patterns visible that a single event never shows. Repeated short openings, none of them long enough to be worth an alert on its own, accumulate into a picture of how a door is being used through a service. Cumulative open time across a shift or across a day is a different figure again from the longest single opening, and it is often the one that explains a temperature that never quite settled.

Temperature is reported at intervals from inside the space and read alongside those door events rather than separately from them. An abnormal rate of rise is detectable as a rate, and because the door record sits on the same timeline, temperature movement can be set against what the door was doing while it moved.

That combination is what produces the substantive result on the refrigeration side, which is diagnostic separation. Reporting distinguishes refrigeration equipment failure from a door left open, from repeated staff access, from delivery and stock-loading activity, from a door contact or seal problem, from sensor failure, and from communications failure. Without that separation all seven of those look the same on a temperature chart: a line that went up. Each has a different response, a different cost and a different person who needs to know, and telling them apart before anyone attends is worth more than the temperature reading that started it.

On the extraction side, a rising pressure drop across the filter assembly indicates increasing filter resistance, read against that system's own clean-filter baseline rather than against a general figure. A falling response indicates reduced extraction performance, which is a different condition and is reported as one. Where a fan-run indication is available, a fan running without the expected pressure response is detectable, and that combination is worth having because it points at the extraction path rather than at the filter.

The air-quality reading at the outlet is reported in parts per million alongside the pressure reading. The parameter it measures is set at engineering design for each extraction system, and is not published here.

On the extraction path this is condition-based extraction and filter monitoring, and servicing on condition rather than to a calendar is the commercial point of it. A filter is attended to because its pressure drop says it has loaded, on that extraction system, against its own baseline. The interval stops being the decision.

On the refrigerated side the equivalent is not a single measurement but the separation of causes. A temperature chart records that something happened. Door state with a duration on it, temperature on the same timeline, and device health underneath both, together record which of several quite different things happened, which is the part a chart alone cannot supply.

Both halves are a visibility and alerting layer, and the limits of that are worth stating plainly. Refrigeration temperature and door state here are monitoring and alerting, not food safety assurance, not a HACCP system and not a due-diligence record. The air-quality measurement describes extraction condition, and is not a detection system of any kind. Nothing here replaces statutory extraction inspection, statutory cleaning or the fire-risk controls that apply to a commercial kitchen, and none of those obligations is affected by it.