Refrigerator temperature mapping

A domestic-style refrigerator can freeze product at the back wall and let it warm in the door. Purpose-built pharmacy units are better, but every unit should be mapped before it stores product.

A practical plan

• Five to nine sensors: each shelf level, back and front, plus the door compartment if it is used and the centre.
• Sampling every one to five minutes; door-opening events are short.
• At least 24 hours, ideally longer, under the loading condition that will be used.
• Include an open-door test if the protocol calls for one, and record how long recovery takes.

Outcome

The result is a shelf-by-shelf statement of which positions stay within 2–8 °C, which are excluded, and where the monitoring probe should live — usually in a buffered form representing product, at or near the warmest acceptable position.

Why small units are worse than large ones

A walk-in cold room has forced circulation and a compressor sized for a room. A domestic-style refrigerator has a cold plate, some convection and a door that is a thin panel between the contents and the room. The result is that small units produce the widest gradients of anything a pharmacy or laboratory owns, and the two extremes are only a few centimetres apart.

The last row is the reason mapping matters here. The monitoring probe is almost always sitting in the best-behaved spot in the cabinet, and the monitoring record therefore says the unit is fine while the door shelf has been running at 10 °C.

Buffering the monitoring probe

A bare air probe in a refrigerator alarms every time the door opens, and a monitoring system that cries wolf several times a day is a system whose alarms nobody reads. The standard fix is to buffer the probe — immerse it in a small vial of glycol, propylene glycol or simply water, or use a probe supplied in a buffer bottle — so that it responds at roughly the rate the product does rather than at the rate the air does.

Two consequences follow. The probe must be buffered during the mapping study too, if the study is to establish where it should live; a buffered probe and a bare mapping sensor measure different things. And the buffering has to be recorded, because a change of buffer volume changes the response time and therefore the alarm behaviour.

Running the study

• Five to nine sensors: each shelf level front and back, the door compartment if it is used, and the centre.
• Log every one to five minutes. Door events last seconds to a minute, and a 15-minute interval will miss them entirely.
• At least 24 hours, preferably 72, under the loading and door-use pattern the unit actually sees — including the morning rush, not a quiet weekend.
• Include defrost cycles if the unit has them. An automatic defrost shows as a periodic rise and has to be assessed against the limits.
• Do not open the door to check on the loggers. That is a door event you have added to your own data.

The output is a shelf-by-shelf statement: which positions stay within 2–8 °C, which are excluded from product storage and physically marked as such, and where the buffered monitoring probe belongs — normally at or near the warmest acceptable position.