The Slow, Silent Drift That Undermines Building Performance

By: Anu Kätkä, Product Line Manager, HVAC & Critical Buildings

Here’s a question worth sitting with: when did you last verify that the sensors in your building are still telling the truth?

Not whether they’re powered on. Not whether they’re returning a signal. But whether the values they report still accurately reflect conditions in the space.

For most building systems, the honest answer is: not recently. And that gap matters more than many building owners, operators and engineers may realize.

Drift is not a failure. That’s what makes it dangerous.

Sensor drift is the gradual deviation of a sensor’s output from its true value over time. It doesn’t announce itself. It won’t trigger a fault alarm or a maintenance flag. The BMS keeps receiving data. The controllers keep responding to it. Everything looks operational.

What changes, slowly and invisibly, is the quality of that data.

A humidity sensor that read accurately at installation might read 3% high after some years in service. A CO2 sensor might show 100 ppm above actual concentration after a few months. These are not dramatic failures. They are quiet biases that push your control loops in the wrong direction, consistently, every cycle, every day.

The consequences depend on the application. In a commercial office, drifting CO2 sensors may lead to unnecessary ventilation and wasted energy. In a pharmaceutical cleanroom or hospital, inaccurate environmental measurements may compromise environmental compliance. In a data center, even a modest temperature drift in the wrong direction can mean running cooling at higher intensity than the load demands, or not enough to protect equipment. The specific impact changes. The underlying mechanism is the same.

Why some sensors drift and others don’t

Drift is not inevitable. It’s a function of sensor design, the measurement technology used, the quality of the sensing element, and the operating environment.

Take humidity measurement as an example. Capacitive humidity sensors work by measuring the change in electrical capacitance of a hygroscopic film as it absorbs or releases water vapor. Over time, that film can be affected by chemical contamination, particulates, or sustained exposure to extreme conditions. In less stable sensing elements, these conditions can alter the baseline response and cause readings to drift.

Sensor design can materially affect long-term stability. Vaisala’s HUMICAP technology has been tested over 12 years of continuous operation in demanding outdoor conditions and remained within specification throughout. That is not simply a short-term accuracy claim; it is a long-term stability claim. For building operators, stability can be the more consequential variable over the lifecycle of an installation.

CO2 measurement tells a similar story. Non-dispersive infrared (NDIR) sensors measure CO2 concentration by detecting how much infrared light is absorbed at the CO2-specific wavelength as it passes through the sample. The quality of the optical components and the calibration approach determine how stable that reading remains over years of operation. Sensor testing across manufacturers has shown that long-term performance can vary considerably: some units may drift outside specification within months, while others remain stable for years. A specification sheet alone does not always reveal that difference.

The calibration trap

The standard response to drift is a calibration schedule. But calibration schedules in practice have a problem: they’re designed around availability, not need.

Annual calibration sounds rigorous. But a sensor that drifts outside specification after six months provides inaccurate data for half the year before anyone checks it. In a large building with hundreds of sensing points, the logistics of maintaining tight calibration cycles across the full install base are formidable. Technician time, access requirements, and recalibration procedures add up quickly.

The better answer is to start with sensors that need recalibration less often, because they drift less. This shifts the problem upstream, where it’s cheaper to solve. It also changes the maintenance economics fundamentally: fewer calibration visits, lower labor cost, less system downtime, and greater confidence in the data between visits.

Probe-based systems with exchangeable sensing elements take this further. When the probe itself can be swapped and sent for calibration off-site, while a freshly calibrated probe goes back in service, facilities can preserve measurement continuity while maintaining calibration coverage. For large facilities, the ability to circulate a pool of calibrated probes through the install base keeps every sensing point current without the constraints of scheduling in-situ calibration for each one individually.

Precision where it counts most

Not every space in a building carries the same measurement stakes. An atrium temperature sensor and a pharmaceutical cold-room temperature sensor are not the same problem, even if they use the same transmitter platform.

Matching sensor accuracy class to application criticality is a basic principle, but one that’s often handled imprecisely in practice. Over-specifying adds cost without benefit. Under-specifying in critical spaces creates risk that shows up much later.

Knowing which parameter in which space drives which control outcome, and then selecting the accuracy level that keeps that control within the required tolerance, is the engineering work that pays off over the life of the system. A sensor with 1% relative humidity accuracy and a sensor with 3% accuracy are both legitimate choices. Choosing between them well means understanding what a 2% humidity error actually costs in your specific application: additional energy consumption, compliance exposure, product-quality concerns or equipment risk.

That judgment requires data. It also requires sensors stable enough that the accuracy class you specify is the accuracy class you get, not just in month one, but in year five and beyond.

The measurement layer is infrastructure

HVAC engineers think carefully about duct sizing, coil selection, and control valve authority. These are hard components with well-understood failure modes and replacement costs. Sensors sit at the edge of that infrastructure, easy to overlook because they’re small, inexpensive relative to major plant, and largely invisible when they’re working.

But the measurement layer is what the entire control system sees. It’s the foundation of every decision the BMS makes. When it’s accurate and stable, the system performs as designed. When it drifts, the gap between the system you designed and the system that’s actually running widens, quietly, until something forces the question.

Build that question into the design process from the start. Ask not just what accuracy the sensor provides at installation, but what accuracy it will provide in three years. Ask how it’s maintained at scale. Ask what the recalibration plan looks like across hundreds of points.

The answers shape system performance throughout the service life of the building.