Cold chain temperature monitoring for delivery fleets

Loggers versus monitoring

A standalone data logger records temperature to internal memory and is read at the end of the trip. It is cheap, it is reliable, and it is evidence. What it cannot do is tell you at 11:04 that the reefer has been climbing for twenty minutes.

Vehicle-based monitoring connects the probe to the tracking device, so the reading is transmitted with the position. That makes three things possible that a logger cannot do: alerting in real time, correlating temperature with location and door state, and producing a record that ties a specific excursion to a specific stop.

Where excursions actually happen

Excursions are rarely caused by refrigeration failure. In practice they cluster around:

• Extended door-open time at a delivery point. The single most common cause. A five-minute drop is recoverable; a twenty-five-minute one at a loading bay in Singapore's ambient conditions frequently is not.
• Pre-cooling skipped. Loading a warm box at the start of the shift, then blaming the unit.
• Engine off during a long stop on vehicles where the refrigeration depends on the engine.
• Overloading past the return-air path, so the readings look fine at the probe and the far end of the load is not.
• Setpoint changed by the driver, deliberately or accidentally.

Notice that four of the five are behavioural, not mechanical. That is why correlating temperature with position, door state and stop duration is more useful than temperature alone.

Why door events matter as much as temperature

A temperature reading without context invites argument. A reading alongside a door-open event, at a known customer site, for a known duration, does not.

Door sensors are inexpensive and change the character of the data. They let you answer the questions that actually arise in a dispute:

• Was the box at temperature when it left the depot?
• How long was the door open at the delivery point?
• Did the excursion begin before or after the door opened?
• How long did recovery take, and was the next drop affected?

The same data supports a conversation with a customer whose receiving process is the actual cause — a conversation that is very difficult to have without evidence.

Designing an alert that gets acted on

Most cold chain alerting fails not because the threshold is wrong but because the alert goes nowhere useful. Three design rules:

1. Alert on trend, not just threshold. By the time you cross the limit you have already lost margin. A rate of rise that will breach in fifteen minutes is the useful trigger.
2. Send it to someone who can act. An email to a shared inbox at 11pm is not an alert. Route to the driver and the supervisor, by a channel they actually read.
3. Suppress the predictable. If every delivery produces a door-open dip, alerting on all of them trains everyone to ignore alerts. Set the threshold above normal operating variation.

Probe placement and calibration

Two practical points that determine whether your data means anything:

Placement. A probe in the return air stream reads the unit's performance. A probe in the load reads what the customer receives. They are not the same number and it is worth knowing which one you are quoting. For multi-compartment vehicles, one probe per compartment, or you are extrapolating.

Calibration. Probes drift. Establish a calibration interval and keep the certificates — if a claim is ever disputed, an uncalibrated probe is an uncorroborated number. Whoever supplies the system should be able to tell you the drift specification and the recommended interval.

What to specify in a tender

• Continuous logging at a stated interval, transmitted with position.
• Number of probes and compartments, and stated placement.
• Door-open sensing on every load compartment, correlated with temperature.
• Alerting on rate of trend as well as absolute threshold, with a named recipient and channel.
• Retention period for temperature records, and export format.
• Behaviour when connectivity is lost: does the device buffer and forward, or is the gap permanent?
• Calibration interval and who is responsible for it.

That last-but-one item is the one most often omitted and most often regretted.