How many sensors are needed for temperature mapping?

The honest answer is that the number of sensors is an output of the study design, not an input to it. Guidance documents give rules of thumb and worked examples, but they consistently ask the study to justify its sensor plan from the space itself.

What drives the count

• Volume and geometry — a small refrigerator and a high-bay warehouse are different problems.
• Known risk points — doors, loading bays, evaporators and air-handling outlets, windows and external walls, ceiling and floor levels, the geometric centre.
• Storage pattern — racking, pallet positions and the height at which product is actually kept.
• Acceptance limits — a tight 2–8 °C range leaves less room for gradients than a 15–25 °C room.
• Whether the study is empty, loaded, or both.

A defensible approach

Start from a three-dimensional grid that covers the storage volume at more than one height, then add sensors at each identified risk point. Document why each location was chosen. Use a sampling interval short enough to capture door-opening events and defrost cycles. Position sensors so they measure air where product sits rather than surfaces or airflow directly from a diffuser.

A common trap is to size the study to the number of loggers available. If the plan calls for more sensors than you own, run the study in phases or hire loggers — but keep a set of common reference positions across phases so the datasets can be compared.

What the analysis needs from the plan

Whatever the count, the analysis depends on clean metadata: a unique sensor identifier, a described location, and the sensor's calibration record. ValiTrac's pre-flight step checks that these are present before any statistics are computed, and it refuses to guess the temperature unit from the values.