Temperature Mapping Sensor Placement: How Many Sensors Do You Need?

As a widely used rule of thumb, a small chamber or refrigerator needs a minimum of around 9 sensors, a cold room typically needs 9 to 15, and a large warehouse needs anywhere from about 20 to more than 100, with the exact number driven by the volume, shape and complexity of the space rather than a single fixed formula. The number is a means to an end: you need enough calibrated loggers to characterise the whole three-dimensional volume and to deliberately over-sample the points most likely to fail. Under-sample and you miss a hot spot; the count exists to make sure you do not. The figures below are practical starting points that a scoped protocol then refines for your specific facility.

Sensor counts by space type

These are common starting points, not regulatory minimums. The protocol always adjusts them for the real geometry of the space:

• Small refrigerator or benchtop chamber: a minimum of about 9 loggers, arranged as one at each of the 8 corners plus one in the geometric centre. Very small units may use fewer, but 9 is a sensible floor for a unit storing regulated product.
• Reach-in fridge, freezer or stability chamber: roughly 9 to 15, adding mid-shelf and door-side points to the corner-and-centre pattern.
• Walk-in cold room: typically 9 to 15 for a modest room, scaling up with floor area and rack height. Multi-level racking pushes the count up because each level needs coverage.
• Large warehouse or distribution store: from about 20 into the dozens or beyond, driven by floor area, ceiling height, number of aisles and the number of doors and docks.

For a large open space, one common approach is to divide the floor into a grid of zones and place a vertical stack of sensors in each zone, so that both the plan and the height of the space are covered systematically.

Think in three dimensions, not two

The single most common placement mistake is mapping only the floor plan. Temperature stratifies vertically: warm air rises, so in a cooled space the ceiling is usually the worst-case hot spot, while in a heated space the floor is coldest. A grid that covers the floor beautifully but sits all its sensors at chest height will completely miss the ceiling hot spot where product on the top racks actually sits. Always build the grid at multiple heights: low (near the floor), middle (working height), and high (near the ceiling or the top of the highest rack). For tall warehouse racking, that may mean three or more vertical levels per grid position.

The worst-case points that must always be covered

Beyond the even grid, a mapping study should deliberately add loggers at the points most likely to be out of range. These worst-case positions are where excursions actually happen, and finding them is the entire point of the exercise:

• Doors, loading docks and air curtains: every opening leaks conditioned air and lets in Singapore's hot, humid outside air.
• Directly in and directly out of the evaporator or cooling air stream: the coldest point is often right at the coil outlet, and a warmer stagnant point sits where the air does not reach.
• External and sun-exposed walls and the roof line: in Singapore, solar and ambient heat gain through the building envelope is a real and continuous load.
• Ceilings and the tops of high racks: the usual hot spot in any cooled store.
• Corners and dead zones with poor airflow: stagnant air holds heat or cold longer than the moving bulk.
• Near heat sources: light fittings, motors, battery-charging areas and any equipment that runs warm.
• Beside your existing permanent monitoring sensor: so you can check whether your control system reads the same as the mapped reality.

Why over-sampling the extremes is the goal, not padding the count

It is tempting to think of sensor count as a budget line to minimise. In reality the count is your insurance against a missed excursion. Every corner or ceiling point you leave uncovered is a place where product could sit out of range with no evidence either way. The economical study is not the one with the fewest loggers; it is the one that covers the true worst cases so that the resulting qualification is defensible and your permanent monitoring sensors can be placed with confidence. If you are unsure how the counting works out for your space, the process is laid out in our step-by-step warehouse and cold room mapping guide.

Every logger must be calibrated and traceable

Sensor count means nothing if the sensors cannot be trusted. Each logger in the grid must carry a current calibration certificate traceable to national standards. At Unitest the mapping loggers are calibrated under our SAC-SINGLAS accreditation, LA-2023-0845-C, so the measurement chain is accredited and traceable, while the mapping study itself remains a documented service executed to WHO TRS 961 Annex 9 and its Supplement 8, and to HSA GDP (GUIDE-MQA-013) or GMP protocol. Where a chamber's own performance is being confirmed and mapped, the methods of IEC 60068-3-5 and IEC 60068-3-11 provide the recognised basis. If any of your own loggers are approaching their calibration due date, our temperature calibration service can bring them back into traceability, and where humidity is mapped as well, our humidity calibration service covers the RH channels.

How sensor count and study duration work together

Sensor count is one half of the coverage equation; study duration is the other. A dense grid recorded for only a few hours can still miss a risk that only appears during the busy afternoon or the overnight defrost cycle, while a sparse grid recorded for a week will faithfully log the wrong points for a long time. The two need to be scoped together. As a rule, the larger and more operationally variable the space, the more you gain from both more sensors and a longer run. A small stability chamber with steady conditions might be well characterised by nine loggers over 24 to 72 hours. A busy distribution warehouse with many doors and shift patterns needs both a larger grid and a full working week, so that every sensor captures the daily and weekly cycle at its own location. Think of count as spatial coverage and duration as temporal coverage: a defensible study needs enough of both.

Adjusting the count for real-world complexity

The starting figures assume a reasonably regular, rectangular space. Real facilities rarely oblige, and several features push the sensor count up:

• High ceilings and tall racking add vertical levels, and each level needs its own ring of sensors.
• Multiple doors, docks or air curtains each create a local disturbance that deserves its own logger.
• Irregular shapes, mezzanines and partitions break the simple grid, so extra sensors are needed to cover each sub-region.
• Multiple cooling units mean multiple air streams and multiple potential cold and stagnant zones to sample.
• Mixed storage classes in one space (for example a chilled zone within an ambient store) each need enough coverage to qualify independently.

None of these is a reason to pad the count arbitrarily. Each is a specific, justifiable reason to add sensors where the risk genuinely lives, which is exactly what a written protocol documents.

A worked example of counting for a space

Consider a modest walk-in cold room, roughly four metres by three metres, with two-level racking and a single door. A defensible grid might run as follows: nine positions across the floor plan (a three-by-three grid), each sampled at two heights to cover the low and high racks, giving eighteen loggers. To that you add worst-case points: one at the door, one directly in the evaporator air stream, one at the return-air path near the ceiling, and one beside the existing monitoring probe. That takes the study to around twenty-two loggers for a room that a naive count might have tackled with nine. The extra loggers are not waste; each one covers a specific place the room could fail, and together they turn a rough guess into an audit-ready qualification you can defend in front of an auditor who asks why a particular corner was, or was not, included in the study.

Placement checklist

• Cover the full floor plan with an even grid of zones.
• Stack sensors at low, middle and high levels in each zone.
• Add loggers at every door, dock and air curtain.
• Cover the coil air stream, both inlet and outlet.
• Cover external walls, roof line, ceilings and rack tops.
• Cover stagnant corners and any heat sources.
• Place one logger beside the permanent monitoring sensor.
• Confirm every logger has a current calibration certificate before the study starts.

Get the count right for your facility

A round number from a table is a starting point, not a protocol. The right sensor count for your space depends on its exact volume, height, layout and the number of doors and heat sources, and it should be set in a written protocol before the study begins. Our temperature mapping service in Singapore and temperature and humidity mapping service scope the logger count and placement for warehouses, cold rooms and chambers. To get a placement plan designed for your facility, request a scoped quote.