Forty minutes out of range, and nobody knew until it arrived

A refrigerated truck leaves the depot at five with a load that has to stay between two and eight degrees. At twenty to eight it is stuck in traffic, and the driver shuts the engine off to save fuel while he waits. The refrigeration unit stops. Forty minutes later it restarts, and the box is back in range well before delivery.

At the store, the thermometer reads five degrees. Everything looks fine. Nobody knows what happened on the road, and nobody knows the load touched fourteen degrees.

The problem is not the measurement points. It is the space between them

Traditional cold chain control measures at points: leaving the depot, arriving at the destination, sometimes a logger in the box that gets read afterwards.

Point measurement answers what the temperature was then. It says nothing about what happened in between — and in between is precisely where product is lost.

For food, the loss is often shelf life that quietly shortens with nobody able to say why. For vaccines and some pharmaceuticals, a single excursion invalidates the batch, however normal it looks on arrival.

What continuous monitoring changes

Seeing it now rather than afterwards. Sensors reporting every few minutes over a mobile network let a control room see which vehicle is in trouble while it is still on the road. Calling the driver at minute five is a different business from finding out at the delivery bay.

Evidence you can hand a customer. Food producers and pharmaceutical distributors supplying hospitals or major retailers have to demonstrate the chain held. A temperature trace for the whole journey settles a dispute in a minute.

Systematic root cause. Once there are months of data, patterns surface on their own: which vehicle swings more than the others, which route has repeated trouble at the same time of day, which delivery point has doors held open far longer than the rest. None of that is visible in a single incident.

Mistakes that recur at installation

Sensors in the wrong place. A sensor near the cold air outlet always reads well, while the pallet at the back by the door runs several degrees warmer. The place to measure is the worst position, not the easiest one to mount.

Alerts tuned too tight. A system that alarms whenever a reading touches a limit will send dozens of messages a day from ordinary door openings, and within a week everyone will have muted it. Good alerting reasons about duration, not just about a threshold.

No named owner. An alert going into a group chat with thirty people in it is an alert nobody owns. Decide who receives it, who acknowledges it, and where it escalates if nobody does within a set number of minutes.

Start with one route

Fit sensors to two or three vehicles on whichever route causes the most trouble, gather a month of data, and see what you did not know. There is almost always something, and it is usually not what anyone expected at the start.

What GIPSIC does here

IoT work does not end at reading a sensor. It is choosing hardware that survives real conditions, designing what happens to readings when the signal drops, storing a great deal of data so it can be searched back quickly, and installing on site with people who know where to mount a box so it does not come loose when the truck hits a pothole.

We design the devices, write the firmware, build the cloud platform behind them, and have our own installation teams in Bangkok, Chiang Mai, Phuket and Nakhon Ratchasima. The whole chain sits in one team.

If you would like to talk about your cold chain, get in touch.