Laboratory refrigerator and freezer validation

Almost every laboratory has a 2–8 °C refrigerator holding something that matters: reagents, reference materials, vaccines, samples. It is also the unit most likely to have a validation file that conflates three separate activities, which becomes obvious the moment an auditor asks which document shows that the bottom shelf stays above 2 °C.

Three activities, three questions

The common failure is a file containing a calibration certificate for the built-in display and nothing else, offered as validation. It establishes that one sensor in one position reads correctly. It says nothing about the other shelves, nothing about recovery after a door opening, and nothing about what happens when the compressor fails.

Mapping a small unit

Volume-based sensor counts developed for warehouses give absurd answers for a 400-litre cabinet, where the practical pattern is the eight corners of the usable volume, the geometric centre, and one sensor adjacent to the unit's own control or display probe so the two can be compared. Add a position inside any drawer or crisper that will hold product, because those are frequently the warmest places in the cabinet and are routinely left out.

Run it loaded, with the load you actually use, and run it long enough to capture several compressor cycles. For a freezer, long enough to capture a defrost cycle — which for an auto-defrost unit may be every six to twelve hours, and is the single most likely source of an excursion. A twelve-hour study on a unit that defrosts every twenty-four hours can produce a flawless report about a cabinet that routinely goes out of specification.

Domestic refrigerators are built to keep food cold, which means cycling the air temperature widely around a comfortable average and defrosting whenever convenient. They are a poor choice for anything with a 2–8 °C requirement, and no amount of mapping makes a unit stable that is not. Mapping will tell you that honestly; it will not fix it.

The ballast bottle decision

Routine monitoring of a fridge is usually done with the probe in a bottle of glycol or water rather than in open air, and this is a deliberate choice with a consequence worth understanding.

• A bare air probe responds in seconds. It will record every door opening as an excursion, because the air genuinely does warm within seconds, and your alarm will cry wolf several times a day.
• A probe in a 30 ml glycol bottle has a thermal time constant of minutes to tens of minutes. It approximates what a vial of liquid product experiences, so it ignores brief door openings and reports the condition that matters to the stored material.
• Mapping, by contrast, usually wants the air: the question is where the space is warm, and a damped sensor hides the gradients you are looking for.

So the two are not in conflict — they are answering different questions, and the protocol should say which sensor type was used for which purpose. A monitoring system validated with a ballast bottle and then compared against a mapping study done in air will show discrepancies that are entirely expected and look alarming if nobody wrote down why.

Alarms, and the number people forget

Alarm limits are not the storage limits. If the product requires 2–8 °C and the alarm is set at 2 °C and 8 °C, every alarm is already a deviation. Alarm limits belong inside the storage limits by a margin derived from the mapping — the observed spread plus the measurement uncertainty — so the alarm fires while there is still time to act.

The second number is the delay. An alarm with no delay fires on every door opening; an alarm with too long a delay fires after the product is compromised. The recovery data from mapping is what sets it: if the cabinet returns to range within four minutes of a normal door opening, a five-minute delay suppresses the noise without hiding a real fault.

Requalification

On a periodic interval you can justify, and on change: compressor or thermostat repair, relocation, a change of load pattern, or a change to the monitoring system. A unit moved to a different room has not been qualified where it now stands — ambient temperature is one of the inputs, and a cabinet that held 2–8 °C in a cool corridor may not in a warm laboratory.