IoT Sensor Placement Guide: Right Sensor, Right Place

The most expensive sensor is the one installed in the wrong place - not because the hardware costs more, but because its data is trusted and misleading at the same time. A temperature sensor mounted in a sunlit corner will drive a cooling system to satisfy a point that represents no one. A vibration sensor bolted to a flanged joint will report the pipe’s resonance rather than the pump’s health. Placement is the invisible variable that determines whether a sensing program produces decisions or decorations, and it is worth more attention than the choice of sensor model.

Placement Is a Sampling Problem

Every sensor samples a larger physical field, and the art of placement is choosing a location where that sample represents what you actually care about. Three concepts govern most decisions:

• The measurement point must match the physical quantity you want: air temperature, not radiant heat; bearing vibration, not housing flex.
• The spatial average matters more than the point value for control. A room’s temperature is not one number; it is a gradient from floor to ceiling, window to core, supply diffuser to return grille.
• The temporal character matters too - a sensor that captures the transient is different from one that captures the steady state.

HVAC Sensing: Representing the Space, Not the Obstruction

Temperature and humidity sensing for HVAC control follows a few rules that are violated constantly:

• Mount away from influence. Avoid exterior walls, direct sun, supply diffusers, windows, and heat-generating equipment. A sensor 30 centimeters from a vent is measuring the vent, not the room.
• Mount at occupied height. Roughly 1.2 to 1.8 meters above the floor. Ceiling-mounted sensors read the stratified warm layer, which can differ from the occupied zone by several degrees in winter.
• Respect the return air. Return-air sensors integrate the room but pick up duct and plenum effects; they are a good trend source but a poor comfort setpoint source.
• Cover the zone’s extremes. For a room with a window wall and an interior core, one sensor cannot represent both. Place at least one near the worst-case load and understand the zone’s response time.

The practical question is always the same: which location would mislead a controller into doing the wrong thing? That location is the one to avoid.

Power and Energy Sensing: Measure the Circuit, Not the Incident

Energy sensors are more forgiving about exact position but less forgiving about electrical context:

• CT placement on the correct phase and load. Current transformers must be on the conductor carrying the load you intend to measure, with correct orientation. A reversed CT produces readings that are wrong in ways that look plausible.
• Beware of paralleled and shared neutrals. A single neutral serving multiple phases must not be measured as if it were a phase.
• Measure at the boundary that matches the decision. Submetering should follow the same boundaries as the cost or control decisions - one meter per tenant, per production line, per chiller - or the data cannot be aggregated into anything actionable.

At panel level, instrument every breaker that feeds a significant, variable load and skip the fixed loads that never change.

Environmental Monitoring: Density by Risk

The right sensor density is a function of the consequence of missing an event, not the square footage.

Space type	Sensing priority	Practical density
Cold storage / walk-in	Temperature, door status	One per zone or room, plus alarm probes
Server room / data hall	Temperature, humidity	One per hot/cold aisle or every 3-5 racks
Museum / archive	Temperature, humidity	One per microclimate zone
Office floor	Temperature, CO2, occupancy	One per HVAC zone (5-10 per floor typical)

For cold chain the rule is strict: a single probe per room misses the warm spot near a door, so use a probe at the evaporator return plus one in the highest-risk location (near the entry or a defrost-heavy section), with door-open events logged as proxy alarms.

Vibration Sensing: The Mounting Is the Measurement

Vibration sensing is where the mounting matters more than the location:

• Mount on the bearing housing or the closest rigid surface. Vibration travels through metal efficiently but loses fidelity through rubber, gaskets, and soft mounts.
• Keep mounting stiffness high. A stud mount beats a magnetic mount; a magnetic mount beats a sticky pad. Each weaker coupling adds its own resonance and attenuation.
• Use consistent orientation across a fleet. Comparing machines only works if sensors are mounted the same way on the same surface; otherwise the differences are artifacts, not faults.
• Match measurement axis to the fault. Most rotating faults show their strongest signature radially; capture at least one radial and one axial axis.

The same sensor on a poorly chosen mount will report false alarms and miss real ones. The mounting is not an installation detail - it is half the measurement.

Coverage versus Cost: The Optimization Problem

Sensor placement is ultimately a budget allocation problem:

1. Start with the decision. List the operational decisions that need data: which zones to schedule, which loads to trim, which assets to watch.
2. Map the minimal set. Place one sensor where it captures the dominant behavior of each zone or asset.
3. Find the second sensor that adds the most. Usually the one that captures the worst-case extreme or the largest untracked load.
4. Add only where the decision improves. Every additional sensor must change a decision, not enrich a dashboard.

A typical office building needs far fewer sensors than intuition suggests - one well-placed sensor per HVAC zone plus a few strategic sub-meters usually covers the decisions that matter.

Verification Before You Trust

Placement errors are easiest to catch in the first week:

• Compare against a reference - a handheld calibrated instrument at the same point.
• Check the trend - does the sensor respond to known events (door opens, compressor starts)?
• Confirm the range - does it read within the plausible operating band for its location?
• Catch the classic mistakes - sensors behind furniture, above hot equipment, in direct sun, or in dead air.

Conclusion

Sensor placement is where the physics of the building and the economics of the budget meet. The well-placed sensor is cheaper than the misplaced one, because it produces decisions that save energy instead of confusion that quietly costs it. Getting placement right - measuring the space rather than the obstruction, mounting vibration sensors as designed, adding density only where risk demands it - is the highest-leverage work in any sensing program, and it is available before the first device is ordered.

Integrar IoT’s deployment methodology includes placement guidance and verification checks for every sensor class, so the data entering an analytics platform represents the spaces and assets it is meant to represent.