What industrial safety demands of NDIR
Gas measurement in industrial settings is not simply about displaying a number. It informs decisions about whether workers may enter, whether more ventilation is needed, and whether a process must be stopped. A sensor must distinguish the target gas, respond sufficiently quickly to concentration changes, and provide interpretable values despite changes in temperature, pressure, humidity, vibration, and contamination. The fact that power is on must also be distinguished from the fact that a measurement is reliable. NDIR—non-dispersive infrared measurement—has drawn attention in industrial safety because its physical measurement principle fits these demands well.
NDIR does not rely solely on a chemical reaction between a gas and a consumable reagent. It reads the phenomenon in which target molecules absorb energy at particular infrared wavelengths. With an appropriate wavelength and optical-path design, it can therefore continuously measure gases with distinctive infrared-absorption characteristics, such as CO₂. But this advantage is only a property of the sensor module itself. To become an industrial-safety instrument, the light source, detector, optical cell, filters, compensation algorithms, housing, power supply, outputs, alarms, diagnostics, and calibration procedures must be validated as one system.
How NDIR reads concentration
An NDIR sensor consists fundamentally of an infrared light source, a measurement cell through which gas passes, an optical filter that passes only selected wavelengths, and an infrared detector. As light from the source travels through a cell of fixed length to the detector, target gas in the cell absorbs part of the relevant wavelength. Absorption intensity, optical-path length, and gas concentration all affect the relationship between incident and transmitted light. This is often described by the Beer–Lambert relationship, but actual sensors calculate concentration with a correction model that includes broadened absorption lines, detector response, temperature and pressure, and nonlinear regions.
Non-dispersive means that the full infrared spectrum is not spread out and analyzed in detail with a prism or diffraction grating. Instead, a filter tuned to the target gas’s absorption band selects the required wavelength range. The structure is relatively compact and suitable for continuous measurement, but the filter band and optical-cell design determine selectivity and measurement range. At high concentrations, absorption can approach saturation, changing resolution and linearity, so the range must be selected for the intended use.
Why a reference wavelength is needed
Industrial NDIR often uses a reference signal in addition to the target absorption wavelength, rather than reading the target wavelength only once. The reference wavelength is chosen in a region where the target gas absorbs very little. Comparing the ratio of the absorption signal to the reference signal can partly cancel changes that affect both signals, such as light-source ageing, changes in detector sensitivity, and gradual contamination of optical surfaces. Vaisala’s description of its dual-wavelength design likewise states that it alternately measures the CO₂ absorption wavelength and a non-absorbing reference wavelength along the same optical path.
A reference channel, however, is not a universal self-calibration device. Ratio correction may be insufficient if contamination affects the two wavelengths differently, if a filter selectively blocks only the target wavelength, or if condensate changes the optical path irregularly. This is why regular functional checks and verification with gas of known concentration remain necessary even when the reference signal improves stability.
Where CO₂ selectivity comes from
CO₂ molecules exhibit strong absorption characteristics in particular mid-infrared bands. Because NDIR sensors use optical filters targeted at these characteristic bands, they achieve a kind of selectivity different from that of heated semiconductor sensors, which respond broadly to many reducing gases. The fact that they do not respond directly to changes in nitrogen or oxygen concentration, which have little absorption at the target wavelength, is also advantageous for CO₂ measurement. This optical selectivity suits applications that need to track CO₂ concentration directly, such as long-term continuous monitoring, ventilation-status checks, and process-leak monitoring.
However, “CO₂-selective” does not mean that nothing other than CO₂ can have any effect. The tails of absorption bands from other gases, including water molecules, can overlap the filter band, and changes in background-gas composition change the density and pressure conditions of a gas mixture. To reduce these effects, manufacturers use narrow-band filters, multiple wavelengths, temperature, pressure, and humidity compensation, and settings for specific background gases. Users should therefore check the data sheet’s list of interference-gas tests and the compensation conditions. The presence of a gas not listed does not justify concluding automatically that its effect is 0.
An industrial field sensor is a package, not a module
Placing a sensor module from a laboratory bench in the field does not by itself make it an industrial-safety instrument. A field package must allow gas to reach the sensor while controlling droplets and particles. Sintered or hydrophobic filters, waterproof and dustproof housings, corrosion-resistant materials, temperature sensors, heaters, pressure-compensation inputs, stable power, and cable glands or industrial connectors are designed together. For fixed equipment, outputs and fault signals such as 4–20 mA, relays, and RS-485 must reach the control system.
A filter is a trade-off between protection and response speed. Fine filters and splash guards reduce the ingress of dust and droplets but slow gas diffusion. Actual product specifications can show different T90 response times depending on the standard filter and spray shield. A pumped configuration can draw a sample to measure a remote point, but tube length, flow rate, leaks, adsorption, condensation, and blockage become new sources of error. Rather than looking only at sensor response time, the total system time for a sample to travel from the inlet to the cell and stabilize must be assessed.
If installed in a hazardous location, the measurement principle and explosion-protection suitability must also be treated as separate questions. It cannot be judged solely from whether CO₂ itself is flammable. It is necessary to consider whether other flammable gases or dust are present at the installation site and whether the equipment’s electrical characteristics, surface temperature, and enclosure meet the area’s requirements. IEC standards for gas detectors address the design, function, performance, and testing of portable and fixed equipment for its intended use. The NDIR principle of a sensor chip cannot substitute for the certification or safety suitability of finished equipment.
Cross-sensitivity and background-gas effects
Cross-sensitivity is the phenomenon in which a component other than the target gas changes a measurement. In NDIR, a typical example is optical interference, where infrared absorption by another molecule overlaps the measurement band. Water vapor affects a broad infrared region and must be considered in environments with large humidity changes. Places containing hydrocarbons, anesthetic gases, refrigerants, or process solvents also require the interference data provided by the sensor manufacturer and tests with the actual gas mixture. A filter design that was adequate in one environment does not guarantee the same performance with a different process composition.
Temperature and pressure matter in addition to optical absorption. The number of molecules in the same volume changes with pressure and temperature, and the output of the light source and detector is also temperature-sensitive. This is why manufacturers incorporate a temperature sensor or accept an external pressure value for compensation. If compensation is switched off or the default pressure value differs greatly from the actual site, the displayed concentration can have systematic bias. Changes in altitude, pressurized ducts, negative-pressure chambers, and warm process exhausts should all be checked before retaining a setting that assumes indoor atmospheric pressure.
Background-gas compensation also requires care. Site air contains oxygen and moisture, while the 0 gas used for calibration is often dry nitrogen. If an instrument is calibrated while applying oxygen and humidity compensation intended for normal air, calibration conditions may no longer match actual measurement conditions. The manufacturer’s procedure must be followed for the compensation settings during calibration and for restoring the original settings afterward. A nitrogen 0 point and an atmospheric-air 0 point must not be treated as the same thing arbitrarily.
Condensation is more difficult than humidity
High relative humidity and water condensed on a sensor surface are different conditions. Even if a data sheet states 0–100 %RH, it may include the limitation “non-condensing conditions.” When temperature drops below the dew point, a water film can form on an optical window or filter, scattering and absorbing light, blocking diffusion paths, and leaving contamination residues even after drying. It can produce a large transient error or a long recovery time, so values during condensation are safer not to regard as normal measurements.
The first response is to understand the dew point and temperature gradient at the installation point. Cold exterior walls, areas near cooling pipes, immediately after washing, and configurations that draw warm, humid gas through a long tube all carry a high condensation risk. Sensor-head heaters, heated sample lines, moisture separators, drainage arrangements, and appropriate flow rates can be used, but each can affect sample composition or response time. Products with heaters must also observe their permitted range and warm-up time; a submerged condition must not be extrapolated as a normal operating condition.
Dust and contamination change measurement slowly
Dust causes problems even though it is not a gas with an infrared absorption band. As it accumulates on a filter surface, it slows diffusion; when it enters an optical cell, it weakens or scatters light. A reference wavelength may partly compensate for common attenuation, but residual error remains if contamination is unevenly distributed or affects particular wavelengths differently. In livestock barns, feed and grain handling areas, cement works and mines, and spaces with combustion dust, it is difficult to apply the same maintenance interval used in a clean indoor environment.
Field management should include visual checks of filter condition, cleaning the enclosure in the manner permitted by the manufacturer, tracking response time, and checking with reference gas. As contamination worsens, the response may slow before the value becomes completely wrong, so changes in the time required to reach a set concentration are also worth recording as an indicator of degradation.
Calibration, bump tests, and functional checks are different
Calibration aligns an instrument’s response to gas at a known traceable concentration with a reference. It generally checks 0 and span and adjusts them if necessary. A bump test is closer to a functional test: it briefly exposes the instrument to gas to confirm that the sensor responds and that the alarm, display, and outputs actually work. Passing a bump test does not prove concentration accuracy, and successful calibration does not mean that the audible alarm and relay are all normal. The two procedures must be managed as having distinct purposes.
When using direct-reading portable gas monitors, OSHA gives priority to manufacturer instructions and emphasizes pre-use functional checks and confirmation of accuracy with test gas of known concentration. NIOSH evaluation guidance considers not only calibration but also linearity, drift, response and recovery time, environmental effects, interference, measurement limits, bias, and precision. An industrial-safety program should not end with one field labeled “most recent calibration date”; it must establish inspection intervals and actions for failures according to each instrument’s risk, frequency of use, exposure environment, manufacturer recommendations, and internal procedures.
The certified value, uncertainty, expiration date, cylinder pressure, composition, and balance gas of calibration gas must also be recorded. ISO 6143 specifies a method for determining and verifying the composition of calibration-gas mixtures by comparison with reference mixtures and for addressing uncertainty. Select a concentration that covers the sensor’s expected alarm point and actual measurement range, and allow sufficient stabilization time. When using a flow-through adapter, observe the flow rate and discharge conditions specified by the manufacturer. Excessive pressure or incorrect flow can change cell pressure or draw in ambient air, producing an incorrect adjustment.
Items easily missed in field calibration
Confirm that the calibration-gas concentration and balance gas suit the sensor and measurement range.
Record the cylinder certified value, uncertainty, expiration date, and values before and after adjustment.
Record the temperature, pressure, and humidity at the calibration site and the instrument’s compensation settings.
Check filters, tubing, pumps, regulators, and adapters for leaks and blockage.
Check not only sensor response, but also the display, alarm, relay, remote transmission, and fault signal.
For equipment outside the allowable error, do not stop after adjustment; examine the cause and the scope of effects on prior measurements.
What NDIR does well—and what it does not
NDIR is strong at continuous measurement of gases with a distinct target infrared band. For CO₂, there is no consumable electrolyte, and long-term stability can be improved with an appropriate reference channel and compensation design. Products can also be configured for high concentration ranges and are readily integrated into fixed transmitters or pumped analyzers. It also differs from combustion-based principles, whose measurement reaction itself stops when oxygen is insufficient. These characteristics are useful in industrial environments that need to monitor CO₂ accumulation, such as confined spaces, fermentation and brewing, refrigeration and food processes, greenhouses, livestock operations, and areas around digestion equipment.
By contrast, an NDIR CO₂ sensor cannot distinguish every cause of oxygen deficiency. CO₂ may read normally even when nitrogen or argon displaces air and lowers oxygen. This is why one CO₂ sensor cannot replace an oxygen sensor, a flammable-gas sensor, and other toxic-gas sensors. In addition, high concentrations beyond the sensor range, optical saturation, rapid temperature or pressure changes, unconfirmed interference gases, condensation, filter blockage, and pump failure can produce falsely low values or slow responses. A display reading “close to 0 ppm” does not always mean the atmosphere is safe.
Alarm settings are also a risk-management decision separate from sensor performance. Legal exposure limits, short-term high-concentration risk, evacuation time, ventilation delay, sensor response time, and installation location must be considered together. Fixing a sensor to the floor solely because of the simple statement that a gas is heavier than air can overlook actual airflow, leak temperature, turbulence, obstructions, and the worker breathing zone. The release source and ventilation flow should be investigated and, where necessary, measurements taken at multiple heights and locations to confirm representativeness.
Adoption decisions start with the use scenario, not the sensor principle
The value of adopting NDIR for industrial safety is determined less by the phrase “accurate sensor” than by how specifically the use scenario has been defined. First, determine which gas to monitor, over what concentration range, how quickly, and at which location. Next, document temperature, pressure, humidity, condensation, dust, and interference gases under normal operation and worst-case conditions. Under those conditions, compare the required measurement range, accuracy, T90, alarms, outputs, enclosure, certification, and calibration method against data sheets and performance-test materials.
IEC 62990-1 distinguishes workplace toxic-gas detection equipment for health monitoring centered on occupational-exposure measurement from safety monitoring centered on alarm operation. This does not mean every CO₂ application automatically falls within the scope of this standard, but its perspective is useful: first define whether the equipment is intended to manage concentration accuracy or to prioritize alarm operation. A purchase specification should state not only the sensor principle, but also allowable error by environmental condition, interference tests, response time, fault indication, calibration interval, component life, filter-replacement conditions, data recording, and maintenance responsibility.
Ultimately, NDIR is not a device that guarantees safety on its own, but one layer in a risk-management system. It should be used alongside ventilation design, work permits, confined-space procedures, independent monitoring for oxygen, flammable gas, and toxic gases, and post-alarm action procedures. NDIR readings become information that can be used for real safety decisions when a physically selective measurement principle, industrial packaging, verifiable calibration, and operation that recognizes field limitations are combined.